Advances in Smart Vehicular Technology, Transportation, Communication and Applications PDF

This book highlights papers presented at the Second International Conference on Smart Vehicular Technology, Transportation, Communication and Applications (VTCA 2018), which was held at Mount Emei, Sichuan Province, China from 25 to 28 October 2018. The conference was co-sponsored by Springer, Southwest Jiaotong University, Fujian University of Technology, Chang’an University, Shandong University of Science and Technology, Fujian Provincial Key Lab of Big Data Mining and Applications, and the National Demonstration Center for Experimental Electronic Information and Electrical Technology Education (Fujian University of Technology). The conference was intended as an international forum for researchers and professionals engaged in all areas of smart vehicular technology, vehicular transportation, vehicular communication, and applications.

121 downloads 3K Views 30MB Size

Report

Download pdf

Recommend Stories

Empty story

Idea Transcript

Smart Innovation, Systems and Technologies 129

Shaoquan Ni Tsu-Yang Wu Tang-Hsien Chang Jeng-Shyang Pan Lakhmi C. Jain Editors

Advances in Smart Vehicular Technology, Transportation, Communication and Applications Proceeding of the Second International Conference on Smart Vehicular Technology, Transportation, Communication and Applications, October 25–28, 2018, Mount Emei, China, Part 1

123

Smart Innovation, Systems and Technologies Volume 129

Series editors Robert James Howlett, Bournemouth University and KES International, Shoreham-by-sea, UK e-mail: [email protected] Lakhmi C. Jain, University of Technology Sydney, Broadway, Australia; University of Canberra, Canberra, Australia; KES International, UK e-mail: [email protected]; [email protected]

The Smart Innovation, Systems and Technologies book series encompasses the topics of knowledge, intelligence, innovation and sustainability. The aim of the series is to make available a platform for the publication of books on all aspects of single and multi-disciplinary research on these themes in order to make the latest results available in a readily-accessible form. Volumes on interdisciplinary research combining two or more of these areas is particularly sought. The series covers systems and paradigms that employ knowledge and intelligence in a broad sense. Its scope is systems having embedded knowledge and intelligence, which may be applied to the solution of world problems in industry, the environment and the community. It also focusses on the knowledge-transfer methodologies and innovation strategies employed to make this happen effectively. The combination of intelligent systems tools and a broad range of applications introduces a need for a synergy of disciplines from science, technology, business and the humanities. The series will include conference proceedings, edited collections, monographs, handbooks, reference books, and other relevant types of book in areas of science and technology where smart systems and technologies can offer innovative solutions. High quality content is an essential feature for all book proposals accepted for the series. It is expected that editors of all accepted volumes will ensure that contributions are subjected to an appropriate level of reviewing process and adhere to KES quality principles.

More information about this series at http://www.springer.com/series/8767

Shaoquan Ni Tsu-Yang Wu Tang-Hsien Chang Jeng-Shyang Pan Lakhmi C. Jain •

•

Editors

Advances in Smart Vehicular Technology, Transportation, Communication and Applications Proceeding of the Second International Conference on Smart Vehicular Technology, Transportation, Communication and Applications, October 25–28, 2018, Mount Emei, China, Part 1

123

Editors Shaoquan Ni School of Transportation and Logistics Southwest Jiaotong University Chengdu, Sichuan, China Tsu-Yang Wu College of Information Science and Engineering Fujian University of Technology Fuzhou, Fujian, China

Jeng-Shyang Pan College of Information Science and Engineering Fujian University of Technology Fuzhou, Fujian, China Lakhmi C. Jain University of Technology Sydney Sydney, NSW, Australia

Tang-Hsien Chang School of Transportation Fujian University of Technology Fuzhou, Fujian, China

ISSN 2190-3018 ISSN 2190-3026 (electronic) Smart Innovation, Systems and Technologies ISBN 978-3-030-04581-4 ISBN 978-3-030-04582-1 (eBook) https://doi.org/10.1007/978-3-030-04582-1 Library of Congress Control Number: 2018961590 © Springer Nature Switzerland AG 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, speciﬁcally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microﬁlms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a speciﬁc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

This volume composes the proceedings of Second International Conference on Smart Vehicular Technology, Transportation, Communication and Applications (VTCA 2018), which is hosted by Fujian University of Technology and is held in Mount Emei, Sichuan Province, China, on October 25–28, 2018. VTCA 2018 is technically co-sponsored by Springer, Southwest Jiaotong University, Fujian University of Technology, Chang’an University, Shandong University of Science and Technology, Fujian Provincial Key Lab of Big Data Mining and Applications, and National Demonstration Center for Experimental Electronic Information and Electrical Technology Education (Fujian University of Technology). It aims to bring together researchers, engineers, and policymakers to discuss the related techniques, to exchange research ideas, and to make friends. Fifty-six regular papers were accepted in this proceeding. We would like to thank the authors for their tremendous contributions. We would also express our sincere appreciation to the reviewers, program committee members, and the local committee members for making this conference successful. Finally, we would like to express our special thanks for the ﬁnancial support from Fujian University of Technology, China, in making VTCA 2018 possible, and also appreciate the great help from Southwest Jiaotong University for locally organizing the conference. September 2018

Shaoquan Ni Tsu-Yang Wu Tang-Hsien Chang Jeng-Shyang Pan Lakhmi C. Jain

v

Organizing Committee

Honorary Chairs Xinhua Jiang Lakhmi C. Jain

Fujian University of Technology, China University of Canberra, Australia; Bournemouth University, UK

Advisory Chair Xin Tong

Fujian University of Technology, China

Conference Chairs Shaoquan Ni Yong Zhao Ruimin Li Xinguo Jiang Wanjing Ma Yaojan Wu Tang-Hsien Chang

Southwest Jiaotong University, China Fujian Normal University, China Tsinghua University, China Southwest Jiaotong University, China Tongji University, China University of Arizona, USA Fujian University of Technology, China

Program Chair Jeng-Shyang Pan

Fujian University of Technology, China

Publication Chairs Tsu-Yang Wu Trong-The Nguyen

Fujian University of Technology, China Hai Phong Private University, Vietnam

vii

viii

Organizing Committee

Local Organization Chairs Huoming Shen Zengan Gao Yong Zhang

Southwest Jiaotong University, China Southwest Jiaotong University, China Southwest Jiaotong University, China

Finance Chair Jui-Fang Chang

National Kaohsiung University of Science and Technology, Taiwan

Program Committees Chien-Ming Chen Dingjun Chen Shu-Chuan Chu Cheng Han

Xingjian Huang Lingming Jiang

Zongping Li Lyuchao Liao Hunton Lin Jerry Chun-Wei Lin Jun Liu

Lan Liu Ling Liu

Ying-Chih Lu Hongxia Lv Si Ma Jinshan Pan Pei-Wei Tsai Yuh-Ming Tseng

Harbin Institute of Technology, Shenzhen, China Southwest Jiaotong University, China Flinders University, Australia Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Southwest Jiaotong University, China Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Jiaotong University, China Fujian University of Technology, China Cheng Shiu University, China Western Norway University of Applied Sciences, Norway Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Southwest Jiaotong University, China Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Fujian University of Technology, China Southwest Jiaotong University, China Southwest Jiaotong University, China Southwest Jiaotong University, China Swinburne University of Technology, Australia National Changhua University of Education, Taiwan

Organizing Committee

Hao Wu

Jimmy Min-Tai Wu Lin Xu Bo Zhang

Guanyuan Zhang Jie Zhang Qiangfeng Zhang Fu-Min Zou

ix

Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Shandong University of Science and Technology, China Fujian Normal University, Fuqing, China Beijing National Railway Research & Design Institute of Signal & Communication Co., Ltd., China Southwest Jiaotong University, China Southwest Jiaotong University, China Southwest Jiaotong University, China Fujian University of Technology, China

Contents

Intelligent Transportation Systems Study on the Algorithm for Train Headway Based on the Simulation of Operation by Driver of High-Speed Railway . . . . . . . . . . . . . . . . . . . Tianyi Sheng, Zhiqiang Tian, and Siyuan Qu

3

The Connection Mode Between Inter-city Rail Transit and Urban Transportation System in Urban Agglomeration . . . . . . . . . . . . . . . . . . Ke Zuo, Lan Liu, and Wei-Ke Lu

11

Research on Regional Transportation Network Development Strategies Based on Regional Synergy: A Case Study of Guangdong-Hong Kong-Macao Greater Bay Area . . . . . . . . . . . . . . . Shunyu Yao and Lan Liu

22

The Risk Attitude Research on Route Choice Problem in Regional Rail Transit Based on Risk Decision Theory . . . . . . . . . . . . . . . . . . . . . Yan Zhang, Hui Zhang, and Bing Wang

29

The Topological Structure of Chengdu Metro Network Based on Complex Network Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . Feng Xue and Chuan-Lei He

37

Key Node Identiﬁcation Method of Chengdu Metro Network Based on Comprehensive Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . Feng Xue, Chuan-Lei He, Zong-Sheng Sun, and Xiao Yu

48

Research on Freights Organization Strategy Based on China-Laos Railway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yihan Wang, Lan Liu, and Hao Huang

59

Classiﬁcation of High-Speed Railway Network Transfer Nodes Based on the Improved Gray Whitenization Weight Function Clustering Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jieying Jiang, Lin Wang, and Si Ma

67

xi

xii

Contents

Research on High-Speed Railway Network Effectiveness Based on Theory of Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Su Liu, Jieying Jiang, and Si Ma

76

Study on the Optimization of Multi-transportation Modes in the Surrounding Area of Subway . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ya-long Lao, Lan Liu, and Kai-yu Yang

85

Coordination Evaluation Model of Metropolitan Rail Transit and Urban Transportation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yi-Fan Yu, Jian-Nan Mao, Lan Liu, and Xue-Jiao Xie

96

Identiﬁcation of Key Nodes and Edges by Importance Ranking and Robustness of Regional Rail Transit Network . . . . . . . . . . . . . . . . . 106 Si Ma and Ruyi Shen The Analysis Method of Regional Railway Network Capacity Loss Under Emergent Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Pu Wang, Yingjie Wang, and Pei-fen Pan Technical Measures of Controlling Train Headway on High-Speed Railway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Yuhua Yang, Shaoquan Ni, Minghui Wang, and Guangyuan Zhang Signal Timing Optimization of Isolated Intersection for Mixed Trafﬁc Flow in Hanoi City of Vietnam Using VISSIM . . . . . . . . . . . . . . 133 Xuan-Can Vuong, Rui-Fang Mou, Hoang-Son Nguyen, and Trong-Thuat Vu Railway Timetable Diagnostic Analysis Based on Train Operation Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 Changan Xu, Shaoquan Ni, Shengdong Li, and Dingjun Chen Research on Train Operation Diagram Compilation of Urban Rail Transit Cross-Line Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Gong Qing Research on the Optimization for the Utilization of Passenger Train Stock of Existing Railway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 Zhiqiang Tian, Zhenbo Yang, Tianyi Sheng, and Rui Zhang The Research on the Key Technologies of the High-Speed Railway Train Diagram Based on Pattern Diagram . . . . . . . . . . . . . . . . . . . . . . 164 Xiaoyuan Lv, Bowen Tian, Xiuyun Guo, Ying Wu, and Huilin Huang Research on the Construction of Emergency Rescue Capability Evaluation Index System for Cross Regional Integrated Transportation Network Under Unexpected Events . . . . . . . . . . . . . . . . 171 Bing Wang, Kang Li, Junjie Li, and Xiuyun Guo

Contents

xiii

Train Transition Timetable Method and Solution . . . . . . . . . . . . . . . . . 177 Lyu Miaomiao, Ni Shaoquan, Jing Huiying, and Geng Jingchun Research on Optimization Technology of Daily Dynamic Train Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Wei Chen and Shaoquan Ni Reversal Headway Analysis for Urban Railway Transit . . . . . . . . . . . . . 193 Hai Zhang, Shaoquan Ni, Changan Xu, Yanqu Cui, and Xueting Li Research on Reliability of Chengdu Rail Transit Network Based on Complex Network Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Wei Chen, Zongping Li, Yi Ai, and Yanni Ju A Time-Based Subway Passengers’ Route Choice Model Using AFC Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 Di Wen, Junjie Li, and Feifei Wei Research on the Construction of Evaluation System of the Hub General Plan of Train Disintegration and Assemblage System . . . . . . . . 216 Boer Deng, Tao Chen, Fugen Shi, and Chunhui Wang Coordinated Control Method of Virtually Coupled Train Formation Based on Multi Agent System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225 Ling Liu, Ping Wang, Bo Zhang, and Wei Wei Coordination Evaluation Index of High-Speed Railway Network Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Tao Chen, Jie-ru Zhang, Hong-xia Lv, and Jin-shan Pan Research on Analysis and Strategy of Clearing Business of Chengdu Metro’s Multi-operating Main Body . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 Yanni Yi, Jinshan Pan, and Xinyi Lin Freight Railroad Service Network Design Problem Considering the Flexible Inbound Time of Shipments . . . . . . . . . . . . . . . . . . . . . . . . 250 Xiaowei Liu, Hongxia Lv, Bin Liu, Liang Xie, and Jingchun Deng Research on Emergency Rescue Plan for Cross Region Comprehensive Transportation Network . . . . . . . . . . . . . . . . . . . . . . . . 259 Shan Huang, Jie Zhang, Hui Zhang, and Changyu Liao Selection Conditions Analysis of Passenger Transport Mode for Fast Passenger Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 Qiangfeng Zhang, Shaoquan Ni, Dong Chen, and Wentao Li Study on Optimal Allocation of Rail Transit Capacity Based on Utility of Passenger Flow Transfer and Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 Qiangfeng Zhang, Shaoquan Ni, Gaoyong Huang, and Wentao Li

xiv

Contents

Connotation and Direction of Urban Rail Transport Integration . . . . . . 298 Tingting Wu, Shuming Huang, Shaoquan Ni, and Rong Kuang Innovation of Networked Railway Transportation Organization in High-Speed Railway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 Shaoquan Ni, Feiyu Yang, and Miaomiao Lv Discussion on the Application of Big Data in Rail Transit Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 Guobao Du, Xuepeng Zhang, and Shaoquan Ni Long and Short Routing Mode of Nanjing Railway Transit Line 3 . . . . 319 Rong Kuang, Wenxian Wang, and Tingting Wu Transfer Optimization of Multi Standard Regional Rail Transit . . . . . . 327 Zhou Xia, Zhang Peng, Lv Hongxia, and Ni Shaoquan Research on the Time Differential Pricing of Intercity Railway Based on Congestion Pricing Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 Zhang Peng, Zhou Xia, Wang Minghui, and Ni Shaoquan Smart Vehicular Technology Research of Fast Image Location Method Based on Improved Sobel . . . 345 Pingjun Zhang, Yang Liu, Yufeng Ji, and Xiaohong Wang Study on Coordination Development Model of the Regional Rail Transit System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 Hao Huang, Lan Liu, Yi-Han Wang, and Jian-Nan Mao Research on Comprehensive Evaluation Method of Regional Railway Network Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360 Feng Zhao, Hong-Xia Lv, and Bing Wang Research on the Development Status of the Heavy-Duty Truck in China and Improving Its Safety and Energy Consumption . . . . . . . . 369 Feifei Wei, Hongye Pan, and lv Hongxia Dynamic Brain Network Evolution in Major Depressive Disorder . . . . . 378 Liping Yang, Yingjie Liu, Bo Zhang, and Hongbo Liu Passenger Centric Timetable Synchronization in Metro Network . . . . . 386 Yuanyuan Wang Brain Structural Abnormalities in Reward and Emotion System in Internet Addiction Disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394 Jinqing Yang, Wei Wang, Zhongyuan Cao, Zhaobin Deng, Wencai Weng, Shigang Feng, Hongbo Liu, and Mingyu Lu

Contents

xv

Research on Optimization of Rescue Resource Scheduling in Inter-regional Integrated Transportation Network Under Emergency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402 Xiaoqian Peng, Xueting Li, Guobao Du, and Jingchun Geng Vehicular Applications and Others The Applied Analysis of Big Data in Trafﬁc Safety Management . . . . . 411 Xiuzhen Yu, Ruifang Mu, Rui Yang, and Lieni Wang An Innovative Design and Simulation of Transom Type Venturi Cooling Design for High-Power LED Headlamp . . . . . . . . . . . . . . . . . . 420 Maw-Tyan Sheen and Qian-ting Wang The Study of Predictive Model for ZD6 Switch Current Based on Wavelet Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430 Junfeng Zheng, Hanyue Zhao, Jie Zhang, and Yunlong Li PCB Image Registration Based on a Priori Threshold SURF Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440 Jing Huang, Junnan Li, Lisang Liu, Kan Luo, Xiaoyong Chen, and Fenqiang Liang The Calculation of Safety Front Boundary of Paired Approach Procedure Based-on Escape Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . 448 X. He, F. Zhang, J. Chen, and F. Song Research About Optimizing the Wake Turbulence Separation for Takeoff of CSPRs Under Crosswind Conditions . . . . . . . . . . . . . . . . 457 Yaqing Chen, Yujie Hou, Dengfeng Hu, and Chunzheng Wang A Generalized FSM Implementation Framework . . . . . . . . . . . . . . . . . . 465 Mao-Hsiung Hung Method on Scale Problem of Passenger Catering Center for High-Speed Railway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471 Geng Jingchun An Improved Linear Population Size Reduction Based Parameters with Adaptive Learning Mechanism Differential Evolution (iLPALMDE) for Real-Parameter Single Objective Black Box Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477 Zhenyu Meng, Jeng-Shyang Pan, Wei-min Zheng, and Xiaoqing Li Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485

Intelligent Transportation Systems

Study on the Algorithm for Train Headway Based on the Simulation of Operation by Driver of High-Speed Railway Tianyi Sheng1, Zhiqiang Tian1(&), and Siyuan Qu2 1 School of Trafﬁc and Transportation, Lanzhou Jiaotong University, Lanzhou 730070, China [email protected], [email protected] 2 Shanghai Railway Administration, the Section of Transportation, Shanghai 200071, China [email protected]

Abstract. The calculation method for train headway of high-speed railway is different from existing railway. Based on the data from EOAS and the CTCS’s Automatic Train Protection system technical characteristics, following these rules, this paper calculates and check the method of headway for high-Speed railway and proposes a method to decrease the interval. Keywords: High-speed railway Train performance calculation Automatic Train Protection Headway

Station and tracking interval are the important parameters of train graph which influence the carrying capacity and the punctuality. The capacity of Beijing-Shanghai HighSpeed Railway which from Xuzhou to Bengbu is rather in shortage at present. It is important for this actuality to reduce the intervals to eliminate the passive position. The train performance calculation for bullet-train is not advanced for the difference between bullet-trains and locomotives which results in relatively large errors now. Tian [1] put forward an algorithm of high-speed railway’s headway based on the CTCS’s characteristic. Chen [2] thought the intervals of trains at a certain station can be reduced based on EOAS. Zhang [3] summarized a relationship between gradient and headway based on his research. Gao [4] proposed that drivers should optimize their driving method to reduce the headway based on the headway of Datong to Xi’an Passenger Dedicated Line. Some scholars [5–7] researched the headways of high-speed railways by method of simulation systems. This paper takes the Tian’s method [1] to research further. In addition, we think it is a fact that drivers’ habit of driving can reflect the influence of constraints from train control system, ramp or the trains’ performance. Based on this we analyzed the composition of headway, the drivers’ habits’ derivation and the optimized algorithm for headway. With the application of big data in railway, the habit can be updated more accurate and painstaking.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 3–10, 2019. https://doi.org/10.1007/978-3-030-04582-1_1

4

T. Sheng et al.

1 Headway and Station Interval of High-Speed Railway It is vivid that the less interval, the higher efﬁciency is. Quasi-moving block system has been applied to all the high-speed railway in China. Train control system takes the realtime circumstance of railway and trainset into account to give a continuous monitoring curve. As Fig. 1 shows, A is the former train and B is the later and the length of tracking distance is Lst. Lst ¼ Ladd þ Lb þ Lp þ Ls þ Lt

B

ð1Þ

A Ladd

Lp

Lb

Ls

Lt

Lst

Fig. 1. The schematic diagram of headway

Where: Ladd is the distance for B’s driver to conﬁrm braking(m); Lb is the distance for B during braking(m); Lp is the additional distance of rain protection system(m); Ls is the length of block section(m); Lt is the length of trainset(m). Therefore, headway Ish can be calculated as this: Ish ¼ 3:6

Lb þ Lp þ Ls þ Lt þ tadd vs

ð2Þ

Where: vst is the speed of train on service(km/h); tadd is the time for B’s driver to conﬁrm braking(s). Other intervals can be analyzed by this method similarly. Successive Route has been abolished for all the lines only bullet-train in service from 2017. Under this circumstance there is no interference between the former and the later train so that all the arrival-departure tracks’ intervals can be uniﬁed as a value. If consider this circumstance that the former train and the later arrived and depart at the same track that the former and the later must be interfered by each other. Only the departure route does been unlocked that the arrival route will be applied. For the conclusion, the headway Ida can be calculated as this formula: Ida ¼ 3:6

Lb þ Lp þ Lin Lt þ Lout arr þ twork þ 3:6 vdep varr

ð3Þ

arr Where: Lout is the length of train to depart the switch fully(m); twork is the time for the station to do some work for the later train(s); Lin is the length of arrival route(m);

Study on the Algorithm for Train Headway Based on the Simulation of Operation

5

vdep is the average speed of the train when departure(km/h); varr is the average speed of the train when arrival(km/h).

2 The Influence of ATP for High-Speed Railway 2.1

Different Strategies Among All Models of ATP

The circumstance of manufacture and application on ATP for rail is in a mess and manufactures published little material for scholars to research it. For CTCS-2 system, there are two types of ATP named 200H and 200C. 300H, 300S and 300T are for CTCS-3 and they are compatible with CTCS-2 also. This paper chooses the 300T which is the worst unfavorable for the efﬁciency under CTCS-3. The drivers of some a railway administration takes B4-rank brake when driving EMU trains equipped with 300T based on the statistics of EMU Engineer Operator Analysis System (EOAS). And the target of the station’s entrance is the switch instead of the entry signal lamp. 2.2

The Drivers’ Operation Under ATP’s Protection

ATP gives the distances of the protection with emergency brake and full common brake and show these in the form of curves including a forecast curve in full supervision mode. Drivers only can observe the forecast curve but they can override it if the speed still under the braking curve. This process as Fig. 2 shows: Speed

EB CB

CB Release

ATP

Emergency Brake Common Brake Forecast Operation

Distance

Driver

Fig. 2. The schematic diagram of the curves calculated by ATP

In all, Drivers should abide by the forecast curve. It is common sense that drivers’ average efﬁciency can approach the limit of protection system’s efﬁciency.

6

T. Sheng et al.

3 The Operation of Drivers Under the Supervision of ATP 3.1

Starting

EMU train is better than locomotives when starting for the advantage that EMU train owns seamless coupler and enough starting tractive force. The process can be dived into two phases, where phase (a) loading weak force to start (This stage is quite short) and phase (b) loading full force. In this paper, it can be considered that the force is fully unloaded in starting stage. The limit speed is 45 km/h when EMU train start at originating station for the partial supervision mode of ATP. The mode will be changed into full supervision after running pass the balise at the end of throat. For the certain length of throat, running at low speed could be exist for a long time during the period. Not only under CTCS-2’s supervision but also under CTCS-3’s must train abide by the process. All the EMU train should start at the partial supervision mode of CTCS-2. When the train running pass the point where the rank of CTCS can be changed, some models of ATP called 300H, 300S and 300T can change into CTCS-3’s supervision. The distance of supervision and ceiling speed of CTCS’s is better than CTCS-2. The ceiling speed can reach to 350 km/h and most of the high-speed railway lines set the ceiling speed as 310 km/h. Starting at intermediate station is contained in this procedure. Starting at originating station’s procedure as Fig. 3 shows. Speed(km/h) 350 300

80 45 0

FS

Throat end

Change Rank of CTCS

Distance

Fig. 3. The schematic diagram of the train departure from originating station

3.2

Braking

3.2.1 Braking to Enter the Station Drivers control the braking to slow down under the supervision of ATP and abide by the limit speed of entrance. After the entry the station, train should stop at the speciﬁed location (normally is the end of the platform). The procedure as Fig. 4 shows. Speed(km/h) 310

80

Start braking

Target

EOA

Distance

Fig. 4. The schematic diagram of the train approaching to station

Study on the Algorithm for Train Headway Based on the Simulation of Operation

7

3.2.2 Braking at Sections Automatic block system has been applied to all the high-speed line. ATP gets block information and determines the end of moving permission with the block changing and the braking curves will be change with it. The algorithm of curves can be referred to the document [8].

4 Examples and Analysis 4.1

Calculation of an Interval of a Certain Station

There is a station’s arrival-departure tracks’ interval relatively larger which is analyzed by the statistics form EOAS. It is necessary to reduce the interval to enrich the capacity. The CRH380B series EMU train equipped with 300T is the worst for efﬁciency and it is widely applied. This paper research this situation. Common braking is divided into 7 ranks generally on all kinds of EMU trains and every rank owns a speciﬁc function to ﬁt it. It is fact that most of the high-speed railway stations are set on the flat ground in this railway administration. Also, the braking distances calculated by ATP will approximate if the gradient from −5‰ to 5‰ [9]. Based on these facts that we assume that this station is set on the ground and own 2 platforms with 6 tracks. This modal calculates the arrival-departure tracks’ interval Ida: 8 < Ida ¼ tin þ twork þ tout coasting tin ¼ tb þ tin þ tstop ð4Þ : coasting tout ¼ tstart þ tout coasting is the Where: tb is the time during braking before entering the station(s); tin time between stop and entry(s); tstop is the time during the stop procedure(s); tstart is the coasting time for train to accelerate to the limit speed of switch(s); tout is the time between having accelerating and leaving the throat(s). The former analysis has conﬁrmed the starting and braking procedures. In addition, when the train approaching the mark of stop, drivers could user lower rank of braking than B4-rank. To simplify the modal, we set a coefﬁcient of B4-rank braking to simplify the process. This modal is similar with the normal model [10] from train performance calculation. But the coefﬁcient of rotary mass is a disturbance for this modal. This modal set this as 116(braking) and 118(starting) instead of 120 from Regulations of Train Hauling Calculation [11, 12]. All the coordinates are showed as Table 1:

Table 1. The coordinates of limited point Entry signal lamp 0

Scissor crossover 50 m

Switch (Lateral) 314 m

Start of the platform 685 m

End of the platform 1135 m

8

T. Sheng et al.

This station applies swingnose crossing which frog angel is 1/18. The limit speed is 80 km/s and we set the train coasts at speed of 77 km/h. The composite force for hauling c1 and B4-rank braking c2 ﬁtted by the data from manufacture is showed as Fig. 5:

Fig. 5. The schematic diagram of the composite force of hauling and braking

Running time can be calculated as this formula: Z t¼ t2

t1

Z dt ¼

v1

v2

v dv nc

ð5Þ

Where: t is running time(s); t1, t2 is the time point of begin and end(s); v1, v2 is the speed (km/h) of the point at begin and end; v is Running speed(km/h); n is the coefﬁcient of rotary mass; c is Composite Force (N/kN). Integral method makes the method hard to solve and cut and try method must be applied at the discontinuous point. The mess make it difﬁcult to solve it by programming. We put the method that use secant to approach the curve to solve it. Every result of micro-unit can be calculated by program iteratively. Modal shows as this: 8 Mv ¼ vi þ 1 vi ¼ 2 km/h > > v2 1 v2i > > < Msi ¼ i þ2nc p i n P > s ¼ Msi > > > i¼0 : i ¼ 0; 1; 2; 3 . . .

ð6Þ

Where: Mv is the interval of unit(km/h); Msi is the braking distance of the unit(m). cpi is the average composite force in the unit(N/kN).

Study on the Algorithm for Train Headway Based on the Simulation of Operation

9

Table 2 shows the parameters and results: Table 2. The Parameters of Arrival the Station coasting t coasting t stop tstart tout work Ida tin 146 s 30 s 60 s 45 s 10 s 36 s 327 s

tb

The gap of result between this modal and traditional modal can reach to 51 s which traditional algorithm gives the result 276 s. The value ranges from 320 s to 337 s is reasonable based on the experiment undertaken by CRH380BL-3531 train at this line. It is vivid that new modal is more superior. 4.2

The Headway After Speed Increase of Beijing to Shanghai HighSpeed Railway

This example set as that double-linked CR400BF train equipped with 300T runs at speed of 350 km/h. Headway Ish can be calculated as this: Ish ¼ 3:6

Lb þ Lp þ Ls þ Lt þ tadd vs

ð7Þ

All parameters and result as Table 3 show: Table 3. The parameter and result of 4.2 Lp Ls Lt tadd vs Lb Ish 110 m 2000 m 420 m 16 s 347 km/h 16600 m 215 s

If no unfavorable condition existing such as electrical sectioning near station or steep slope, the headway can be kept at 4 min. This is an ideal example for the electrical sectioning, steep slope and other unfavorable factors on the railroad midway. In addition, the calculation of curves is not based on the single braking performance of train so it is hard for drivers and us to ﬁt the braking curves. ATP is just a protector instead of indicator, the intelligent operation of China high-speed railways is still fuzzy now. ATO is going to be introduced to EMU trains but it should be protected by the curves of ATO yet. Experienced drivers can perform as well as ATO. The algorithm of ATP or ATO still should be optimized to ﬁt the characteristics of train and drivers.

10

T. Sheng et al.

5 Conclusion It is a contradiction that no enough research on bullet-train performance calculation and cost of experiment is quite high. The result from calculation is unreasonable and useless. The new mode of this paper can balance the mess from the accuracy and low cost. In fact, the station’s interval calculated is 7 min. The experiment and calculation get the same result so the intervals can be declined to 6 min. Parameter tb’s ratio of the interval is quite large which reaches to 44.6%. The utilization ratio of B4-rank braking is approximate 50% so there is quite allowance of braking system. When B7-rank braking applied, tb can be declined to 104 s so Ida can be reduced to 285 s too. The 13% of save is meaningful to the high burden lines. The parameter of ATP should be changed to ﬁt the situation. Acknowledgement. This research was supported by the National Natural Science Foundation of China (Grant No. 71761023, 61863024), Natural Science Foundation of Gansu Province, China (Grant No. 18JR3RA107, 18JR3RA110), and the Young Scholars Science Foundation of Lanzhou Jiaotong University (Grant No. 2016020).

References 1. Tian, C.-H., Zhang, S.-S., Zhang, Y.-S., Jiang, X.-L.: Study on the train headway on automatic block sections of high sped railway. J. China Railw. Soc. 37(10), 1–6 (2015) 2. Hen, J., Xu, X., Peng, X., Shi, J., Qu, S.: Research on departure-arrival train interval of highspeed railway based on EOAS. China Railw. (8), 64–67 (2017) 3. Zhang, Y., Tian, C., Jiang, X., Wang, Y.: Calculation method for train headway of high speed railway. J. China Railw. Soc. 34(5), 120–125 (2013) 4. Gao, H.-M.: Research on tracking interval time of EMU on DaTong-Xi’an passenger dedicated line. CARS (2015) 5. He, Q., Yin, Y.-Z.: Design of traction calculation and emulation system for EMU. Technol. Econ. Areas Commun. 15(4), 82–84 (2013) 6. Zhang, G., Pan, J.-S., Ni, S.-Q., Lv, H.-X.: Study on train traction calculation simulation system. Railw. Comput. Appl. 16(8), 18–20 (2007) 7. Zhang, Y., Zhao, M., Wang, X.-S.: Research on a train operation simulation system under moving automatic block conditions. J. Syst. Simul. 11, 6 (1996) 8. Wei, H.-J.: Research on the calculation for the curve of ATP in cab. Railw. Signalling Commun. Eng. 10(s1), 33–38 (2013) 9. Zuo, Z., Zhao, X., Liu, L.: Study on train operation modes on long heavy down grade of high-speed railway. Railw. Transp. Econ. 39(5), 31–35 (2017) 10. Mao, B.-H.: Train Performance Calculation and Design. China Communication Press, Beijing (2008) 11. Ministry of Railways of the People’s Republic of China. Regulations of Train Hauling Calculation (1998) 12. Influence of rotary mass coefﬁcient in high speed train traction calculation. Railw. Locomotive Car 30(3), 56–59 (2010)

The Connection Mode Between Inter-city Rail Transit and Urban Transportation System in Urban Agglomeration Ke Zuo1, Lan Liu1,2(&), and Wei-Ke Lu1

2

1 School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. Combining with the deﬁnition of urban agglomeration, this paper divides urban agglomeration cities into three categories and summaries the intermodal transport modes of cities. The types of rail transit in urban agglomeration are summarized and divided into three categories: trunk railway, regional rail transit, and urban rail transit; We conduct the study for the convergence mode of the urban trafﬁc system with the railway trunk system and the regional track system, and further analyze the characteristics of several models and their applicable occasions; Finally, both Chengdu and Tokyo network are analyzed as examples. Keywords: Urban agglomeration network Inter-city channel Trunk railway Regional rail transit Connection mode

1 Introduction Urban agglomeration transportation network is a foundation of urban agglomeration transportation system. According to the theory of non-equilibrium development, the urban agglomeration with transportation networks consists of two parts: cities and development axis. Cities are the core of regional economic and are connected in series through the infrastructures [1]. In the study for the convergence of urban transportation networks, the international community pays much attention to the construction of comprehensive transportation hubs and believes that transportation hubs are the link connecting regional transportation and are of great signiﬁcance to the integrated transport networks [2]. Chiaki Surinam stated that it will be necessary to focus on the development of urban rail transit in the major cities, to solve the problem of the ever-increasing size of the city and the increasing trafﬁc demand. He put forward to corresponding suggestions for the

This research was supported by the National Key R&D Program of China (2017YFB1200702). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 11–21, 2019. https://doi.org/10.1007/978-3-030-04582-1_2

12

K. Zuo et al.

optimization of the connection between rail transit and other modes of transportation [3]; Yang Zhijie categorizes the connection mode between regional and urban rail transit network [4]; Jiang Wen established a fuzzy interval DEA model for the selection evaluation of rail transit connection patterns [5]. The research focusing on the convergence problem are mostly based on the single urban transportation network. There is a lack of research on the connection of the urban transportation network and the theoretical results that can be directly applied to the design of the connection between transportation channels and urban transportation networks. This paper takes rail transit as paper object, analyzes the channel convergence pattern between central cities and ordinary cities in urban agglomerations, categorizes the levels of rail transit in urban agglomerations, and elaborates the mode of the trunk railway system and the regional rail transit connecting with the urban trafﬁc system.

2 Inter-city Transportation Connection Model According to the results of the urban nodes division, the inter-city transportation and urban trafﬁc convergence patterns are generally divided into three categories [6]. (1) The Connection Mode between Central City and Central City The convergence model can be summarized as “dumbbell type” between two central cities, which is mainly connected through the middle and long distance transport corridors. Among them, the main channels are highways, high-speed railways, and the National Railways as Fig. 1.

Fig. 1. Connection between Central City and Central City

(2) The Connection Mode between Central City and Sub-center City Due to different city size, highway trafﬁc and rail transit routes diverge in central cities and merge in sub-central cities, whose forming structure is like a “cone”. Among them, the main channels are highways, intercity rail transit, and highways as Fig. 2.

The Connection Mode Between Inter-city Rail Transit

13

Fig. 2. Connection between Central City and Sub-center City

(3) The Connection Mode between Central City and Ordinary City Generally, there are many ordinary cities around the central city. The ordinary city has a smaller volume. Also its convergence with the central city is smaller, and the mode is relatively simple. Among them, the main channel is mainly expressways and inter-city rail transportation as Fig. 3.

Fig. 3. Connection between Central City and Ordinary City

In addition to the three major urban convergence modes mentioned above, the intercity convergence patterns in urban agglomerations are also linked among sub-central cities, sub-central cities and ordinary cities. They are usually connected through the national highway or highway to form a single dot-line convergence mode, Because of the less correlation.

14

K. Zuo et al.

3 Urban Agglomeration Rail Transit Level Division This paper summarizes the urban rail transit as the following ﬁve types (Table 1), considering the rational layout of the rail transit industry and the positioning of transportation functions [7]. Table 1. Division of urban agglomeration rail transit Project

Ordinary railway

Line properties

Mixed flow of both passenger and freight National railway network foundation

Functional position

Service area

Maximum speed

City with more than 200,000 population, border towns, major ports, resource development sites 160 km/h

Passenger dedicated railway line Passenger

Inter-city rail transit

Municipal railway

Urban rail transit

Passenger

Passenger

Passenger

National railway network core

National railway network supplement

Urban public transport backbone line

National megacities, regional center cities, capital cities, transportation hubs

Regional town dense area

The skeleton of municipal trafﬁc; Intercity rail supplement Municipal town dense area

200– 350 km/h

120– 200 km/h

100– 200 km/h

40– 100 km/h

Important municipal groups

The ﬁve kinds of urban agglomeration rail transportation above can be divided into three levels: trunk railway, regional rail transit, and urban rail transit. There is a clear division in passenger transport between trunk railway systems and regional rail transit systems. The trunk railway system mainly undertakes the transportation of cross-boundary passenger flow (trains passing through the area but passengers not getting on and off the train), including ordinary railways and passenger dedicated railway lines; The regional track system is responsible for the passenger flow between major cities or towns inside the urban agglomerations, as well as the passenger flows between some urban groups and sub-center towns, including intercity rail transit and municipal railways. There are great differences between the trunk railway and the regional rail system in terms of transportation organization, service scope and station setting.

The Connection Mode Between Inter-city Rail Transit

15

4 Connection Mode Between Urban Rail Transit and Urban Transportation System 4.1

Connection Between Trunk Railway and Urban Transportation System

The connection between trunk railway systems and urban transportation systems is mainly accomplished through site transfer, after the introduction of the trunk railway line into a city. In the setting of convergence points, different numbers of stations will be set depending on the size of the city, which can be roughly classiﬁed into the following categories: (1) One-station Connection The small and medium-sized cities in the urban agglomeration have a small population and short distances within the city, and there are few railways. Due to the difﬁculty of project implementation, the investment scale, and the impact on the urban spatial structure, there is only one railway passenger station in a city. All trunk railway lines are introduced to the station. In this case, the link between the trunk railway and the urban transportation system is only one-station connection through the railway passenger station. Various transportation modes such as urban rail transit, regular public transportation, taxis, and social vehicles are introduced into the railway station area to form an integrated transportation transfer system. In some cities, due to early insufﬁcient planning considerations and restrictions on geographical conditions, the new lines could not be introduced into old passenger terminals. At this time, the one-station connecting city will expand or build a new passenger station on the original basis. In this circumstance, the connection between the railway trunk line and the urban transportation system was established through the construction of a new railway passenger station to form an integrated transportation and transfer system. (2) Multi-station connection In large cities, a number of railway passenger stations are set up, taking into account the conditions of route introduction and the travel distance of passengers. In this case, the link between the trunk railway line and the urban transport system are some railway passenger stations as Fig. 4.

16

K. Zuo et al.

Fig. 4. Trunk railway lines are introduced into the city through multiple stations

4.2

Connection Between Regional Rail Transit and Urban Transportation System

Regional rail transit mainly serves the passenger transportation between cities and major towns, urban groups and sub-center towns within the urban agglomeration. It has advantages in mobility, transfer, point-to-point and other aspects. Regional rail transit routes along towns and passenger trafﬁc concentration points within the metropolitan area, covers blind spots on the main railway, and gradually forms a network. This section mainly discusses the connection between regional rail transit and urban rail transit. There are three main modes for the connection between regional rail transit systems and urban rail transit lines: (1) Core Crossing The core crossing type refers to the connection mode in which the urban network line enters and crosses the core area of the city, and the multiple lines of the urban network are exchanged to complete the passenger flow exchange as Fig. 5. The regional rail line crosses the core area of the city, which can maximize the direct transportation of passengers. It is more convenient for city-bound passengers to enter and leave the city center and save the transfer time. However, there are problems such as the increase of construction and operation costs as well as the increase of engineering difﬁculty in the regional track crossing the urban area.

The Connection Mode Between Inter-city Rail Transit

17

Fig. 5. The core crossing type

(2) Peripheral Access Peripheral access type refers to the connection mode in which the regional track lines do not enter the core area of the city and the passenger flow exchange is completed through the transfer or direct operation of multiple lines in the urban network as Fig. 6.

Fig. 6. Peripheral access type

18

K. Zuo et al.

The regional transit network and the city’s track network are independent of each other and can flexibly select the location and layout of the linking station. The regional track line does not enter the core area of the city and can adopt more economical laying methods and save a lot of construction and operating costs. However, the directness of some passengers has been affected. (3) Core termination Core termination refers to the transition mode where the regional transit line enters and terminates at one or more interchange hubs in the core area of the city, and the mode is completed by the urban network through transfer stations as Fig. 7.

Fig. 7. Core termination type

The regional transit line terminates at an interchange hub. The hub may include rail transit area lines, city lines, national railways, and even airports. It can provide passengers with a variety of travel options; however, the transfer stations under this model are larger. The complexity of the passenger flow lines in the station increases the difﬁculty of operation and management.

5 Case Analysis 5.1

Case of Trunk Railway Connection

The Chengdu Railway Hub connects 11 railway trunk lines and 5 railway branch lines, forming a star-shaped radiation pattern that relies on the inner ring to connect multiple directions. The inner ring is a passenger loop consisting of the east and west ring roads, and is connected in series with Chengdu North Railway Station, Chengdu East Railway

The Connection Mode Between Inter-city Rail Transit

19

Station, Chengdu South Railway Station and Chengdu West Railway Station. The outer ring is a freight ring line, which is mainly composed of the north ring line and the outer wrap of the truck. It also connects the freight stations such as Dawan, Chengxiang, Xinjin, Longquan and Chengdu North Marshalling Station. The “three main and two auxiliary” layout is formed on the passenger transport system. Chengdu North Railway Station, Chengdu East and Tianfu Station are the main passenger stations, and Chengdu South and Chengdu West are auxiliary passenger stations. The railway layout is as shown in the Fig. 8.

Fig. 8. The railway layout plan of Chengdu Hub

5.2

Case of Regional Rail Transit Connection

A major feature of the Tokyo Metro is that it runs directly with other private railway companies and JR companies. Through direct operation, passengers from the suburbs to the city center do not need to transfer. At the same time, the direct operation avoids the repeated construction and investment in the underground of the city center, and also guarantees the beneﬁt of both companies. The Yamanote Line is the demarcation line between the downtown area and the suburbs of the Tokyo Metropolitan City Cluster as shown in the Fig. 9. It is also the demarcation line between the subway and suburban railways. The main layout of the subway is in the Yamanote Loop, while the suburban railway is laid out outside the Yamanote Loop. The Yamanote Loop plays a role in the function of transmission, transfer and direct communication. It is the artery line of public transportation in Tokyo.

20

K. Zuo et al.

Fig. 9. The location of the Yamanote line in Tokyo metropolitan area

6 Conclusions This paper analyzes the connection mode of intercity rail transit in urban agglomerations. The research content is mainly divided into four parts: The ﬁrst part studies the connection between intercity trafﬁc and urban trafﬁc; In the second part, the rail transit in urban agglomerations is summarized as ordinary railways, passenger-only lines, inter-city rails, city-area railroads, and urban rails; The third part takes the trunk railway and the regional rail transit as the objects, and studies its connection pattern with the urban transportation system; Finally, the trafﬁc networks of Chengdu and Tokyo are taken as examples to study the connection performance. Because of conditional limitations, data analysis has not been performed on various models. Future research may consider in-depth analysis in this respect.

References 1. Wen, J.: Evaluation rail transit connection mode selection based on interval DEA model. J. Transp. Syst. Eng. Inf. Technol. 12(6), 132–136 (2012) 2. Yang, Z.: Analysis of the connection mode between urban and urban rail transit lines. China Transp. Rev. (10), 48–55 (2013) 3. Guo, F.: Study on connection transfer of urban integrated transportation junction. Huazhong Univ. Sci. Technol. (2004)

The Connection Mode Between Inter-city Rail Transit

21

4. Sato, L., Essig, P.: How Tokyo’s subway inspired the pads RER. Jpn. Railw. Transp. Rev., 3–6 (2000) 5. Liu, Z., Gu, B., Sun, S., et al.: Statistics and analysis of urban rail transit lines in 2014 China——express delivery of annual report on urban rail transit II. Urban Mass Transit (2015) 6. Dong, Z., Wu, B., Wang, Y.L., et al.: Research on development characteristic of transportation system of China urban agglomeration. China J. Highway Transp. 24(2), 83–88 (2011) 7. Li, J.-W.: Study on the system paths for transportation infrastructure integrated development of urban agglomeration. Logistics Technol. 4, 3 (2008) 8. Gao, F., Lei, L., Yu, P.: Analysis on the joining mode of urban and regional rail transit. Railw. Transp. Econ. 31(8), 56–58 (2009)

Research on Regional Transportation Network Development Strategies Based on Regional Synergy: A Case Study of Guangdong-Hong Kong-Macao Greater Bay Area Shunyu Yao1 and Lan Liu2(&) 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected]

Abstract. Regional synergy can effectively take the advantages and characteristics of different cities in the area, strengthen the industrial division and cooperation of the region, and optimize resources allocation. Trafﬁc integration and interconnection comes a guarantee for high-efﬁciency linkage and cooperation of region, while, the regional synergetic development also puts forward new requirements for the construction of transport networks. This paper analyzes the situation and characteristics of regional synergetic development, and points out the problems and corresponding development countermeasures of the transportation network. Keywords: Synergetic development Regional transportation network Coordination and cooperation Management and control system Guangdong-Hong Kong-Macao Greater Bay Area (GHM Greater Bay Area)

1 Introduction Regional synergy, the extended meaning of ‘synergy’, is an open system consisting of people, logistics and information flows between cities. It fully takes advantages and characteristics of different cities by exchanging material and energy without the constraints of administrative divisions, optimizes the allocation of development resources and realizes the maximization of overall regional beneﬁts. Guangdong-Hong KongMacao Greater Bay Area (GHM Greater Bay Area) is a successful case of regional synergy. The transport network is the foundation and support for promoting communication and cooperation in GHM Greater Bay Area. Meanwhile, the regional transportation network development is a focal point of regional synergy. The unique spatial organization structure of this region has proposed higher requirements for the development and construction of its transport network. At present, the trafﬁc connection between cities still stays an uneven level, leading to difﬁcult to build a convenient and efﬁcient regional comprehensive transportation system. © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 22–28, 2019. https://doi.org/10.1007/978-3-030-04582-1_3

Research on Regional Transportation Network Development Strategies

23

However the trafﬁc connection between cities still stays an uneven level, leading to difﬁcult to build a convenient and efﬁcient regional comprehensive transportation system. The corresponding development strategies for regional transportation network basing on the development strategies of GHM Greater Bay Area becomes the guarantee of regional synergetic optimization. This paper is organized as follows. In Sect. 2, the regional synergy and urban development trends are analyzed. Section 3 describes characteristics of regional trafﬁc demand. Section 4 provides background information of GHM Greater Bay Area for the case study. Section 5 proposes problems and challenges for such regional transport networks. Some corresponding strategies are discusses in Sect. 6. Finally, Sect. 7 gives a conclusion for this theme.

2 Regional Synergy and Urban Development Trends Generally, cities can be divided into several grades according to the size of the population. As the center, large cities gradually form an urban agglomeration through further synergetic development with the small and medium-sized surrounding cities. Basically, this evolution process can be divided into four stages as Table 1 lists below. Table 1. Regional urban development and the corresponding trafﬁc characteristic Regional urban Spatial development organization stage structure The independent development stage The central city fast developing stage The central city expanding to peripheral cities stage The networking of the regional urban agglomeration stage

Isolated

Regional trafﬁc demand pattern Connection Timeliness Trafﬁc mode intensity requirement Low Low Ordinary roads

Cohesion mode Axial

Axial expansion

Medium

Relatively high

Highways, ferries and Axial railways

Circle expansion

Relatively high

High

Highways, ferries high-speed railways and expressway

Plane

Network integration

High

Very high

Highways, highspeed railways, expressways, fast tracks and multimode links

Network

The close and frequent contact between different cities has accelerated the industrial division and regional cooperation in the course of the development of urban agglomeration, and it also has promoted the establishment of regional synergetic mode. The trafﬁc demand pattern changed signiﬁcantly in each speciﬁc stage.

24

S. Yao and L. Liu

3 Changing Characteristics of Regional Trafﬁc Demand As the regional synergy and cooperation brings the signiﬁcant change of trafﬁc demand, transport interconnection requirements grow with the increasing intensity of regional interaction. Cities eventually establish comprehensive, three-dimensional, multi-channel transport linkages to meet the transport demands of high-frequency, high-intensity. Meanwhile, regional transport integration can promote the further development of regional urban agglomeration. With the formation of urban agglomeration, the deepening of industrial division and regional cooperation, the boundary between cities will be gradually blurred. For passenger transport, a certain percentage has transformed from tourism, public service and etc. to commute as the travel purpose between cities, and there has a tidal similar to the rush hours in the central city. Moreover, logistics freight more targeted, and becomes further related to the regional characteristics. Basing on these, higher requirements have been put forward. Not only the quality and quantity for transportation infrastructures as requirements on hardware, but the convenient transfer among different trafﬁc tools comes the key point, and building an efﬁcient regional trafﬁc control and management system as well.

4 Guangdong-Hong Kong-Macao Greater Bay Area Background The GHM Greater Bay Area refers to the urban agglomerations formed by 9 cities and two special administrative regions. It has a total population of more than 70 million people and GDP of about 10 trillion yuan in past 2017. Recent years, some regional channel and links between cities are constructing to offer more convenient and fast transport service. Table 2 lists all these projects below. Table 2. Major transport projects for 2016–2020 in GHM Greater Bay Area Project Hong Kong-Zhuhai-Macao Bridge Guangzhou Port channel widening project Shenzhen-Zhongshan Bridge

Investment Over 100 billions

Humen Second Bridge Shenmao railway

Approximately 11.2 billions 60 billions

Nansha Port project phase four Nansha railway

Approximately 30 billions 15.2 billions

Sea ﬁlling project in Chang’an New Area

18 billions and 16.8 km2 to ﬁll

Approximately 2.76 billions Over 20 billions

Period Location Completion Hong Kong, Zhuhai, Macao Till year Guangzhou, Shenzhen, 2019 Dongguan Till year Shenzhen, Guangzhou, 2024 Zhongshan Till year Guangzhou, Dongguan 2019 Year Guangzhou, Shenzhen, 2017–2023 Zhongshan Year Guangzhou 2018–2021 Year Guangzhou 2016–2020 Year Dongguan 2016–2020

Research on Regional Transportation Network Development Strategies

25

In view of the transport development situation in 2017, GHM Greater Bay Area has initially established a modern integrated trafﬁc network combining multi-mode transport tools. Meanwhile, although the modern integrated trafﬁc network has been initially built, currently the comprehensive trafﬁc planning for the bay area has not regard the region as a whole, instead, still stays at city level. In addition, the distribution of trafﬁc infrastructures remains imbalanced, and the supply ability can not fully match demand. The coordination mechanism is difﬁcult to ﬁt the new characteristic elements of regional passenger and freight transportation. There are still problems and challenges faced by regional transport construction.

5 Problems and Challenges for Regional Transport Networks 5.1

Lack of Regional Comprehensive Transport Planning to Guide

Though transport planning and network construction in each city becomes more reasonable than before. The overall consideration under regional synergetic development is relatively not enough. The GHM Greater Bay Area currently has no uniﬁed coordination mechanism for transportation issue. This has led to a lack of depth and efﬁciency in communication between cities, resulting in many disputes that are far from the expectations of regional synergetic development. For example, without regional comprehensive planning and coordination, cities tend to build homogeneous transport facilities for their own interests. As a result of vicious competition, the regional overall efﬁciency and service level declines. 5.2

Insufﬁcient Rapid Links and Transport Channels

Accordingly, as a result of regional synergetic development, commuter trafﬁc will appear as a part of transport demand and the regional communication requirement will grow signiﬁcantly with the expansion of central cities. However, it is difﬁcult to match the fast connection need between cities with current regional networks. Figure 1 shows the changes of transport demand intensity. In GHM Greater Bay Area, this issue can mainly be reflected in two aspects. Firstly, the number of trafﬁc routes and the capacity of key channels can not ﬁt trafﬁc demands especially between eastern and western coast. Some recent major transport projects have lagged far behind plans. Secondly, the network layout needs to be improved and the transport structure is partly unreasonable. At present, there is also an imbalance developing situation among various trafﬁc modes. These imbalances can not only have a negative influence on intensive use of land resources, but go against sustainable development. 5.3

Transport Connection and Transfer Stays Imperfect

The connection and transfer between different trafﬁc modes and different cities is not as unblocked as some city inner road system. In GHM Greater Bay Area, the seamless docking of infrastructures such as highways and waterways needs to be strengthened.

26

S. Yao and L. Liu

Fig. 1. Changes of transport demand intensity map in GHM Greater Bay Area

The coordination of the network connection between the central area and peripheral cities is about to improve. The efﬁciency of transfer among various modes has not fully reached the requirement of regional synergetic development for urban agglomeration. 5.4

Low Level of Transport Management and Control

The importance of transport management and control has already widely realized. Despite of that, actual action to improve this part was less than expected. The GHM Greater Bay Area has gathered more than 70% of the highway toll stations in the province; the management and control are incompatible with regional economic and social development. This refers to high density which has a negative impact on the fluency of trafﬁc flows, unreasonable layout, and different charging system in different cities. Due to the lack of communication and coordination, the complex management and control system causes lots of conflicts each year.

6 Corresponding Strategies for Regional Transport Network 6.1

Improve the Transport Planning at Regional Level

The current transport planning lacks overall consideration for area as a whole. It is not reasonable for the spatial characteristics under regional cooperation. Instead, the overall planning needs to be carried out at the regional level. It means that not only regional characteristics, trafﬁc conditions, but also industrial distribution, policy and other factors should be considered as a whole in order to establish the comprehensive transport planning suitable for regional synergetic development. In the GHM Greater Bay Area case, the transport planning of each city stays mutual independence before the synergetic development strategy was proposed. This situation has resulted in a slow progression for the implementation of transport links. To

Research on Regional Transportation Network Development Strategies

27

improve the problem, more consideration from the regional overall perspective can be necessary. The regional planning should be set basing on regional developing goals and characteristics as early as possible. 6.2

Establish a Regional Coordination System

The development of regional transportation cares not only the communication and cooperation between cities, but the coordination of trafﬁc and space, which ensures urban construction and industrial division reasonable through the systematic integrated transport network supporting. A coordination mechanism at different levels is indispensable in that process, so as to strengthen the coordination efﬁciency between cities and the collaboration between transportation and other industry ﬁelds. Furthermore, set and reasonably allocate the special supporting funds for major regional transport projects can be another critical point. 6.3

Continuously Concrete on the Construction of Transport Infrastructure

It has already been mentioned that transport infrastructures remain imbalanced and irrelative from city to city. The road density, highway level, the coverage of fast tracks and the system capacity of network in central city can be generally higher than those peripheral cities. In the case, the direct trafﬁc link between eastern and western coast of Pearl River estuary is only Humen First Bridge. The lack of regional corridor has delayed the growth of regional synergetic development. To improve this part, as mentioned above, the GHM Greater Bay Area will focus on the highway corridor and the critical rail-link (Hong Kong-Zhuhai-Macao Bridge, Shenmao railway and etc.). In short, before the regional transport network becomes an efﬁcient and integrated system, construction of infrastructures, especially the regional corridor, should always be regarded as an important basis for regional synergetic development. 6.4

Build Intelligent Transport Management System

Nowadays, the rapid development of intelligent transport brings advanced tools and methods for management and control, such as “ETC” system. It has signiﬁcantly enhanced the running efﬁciency of the toll stations, especially on rush hours. Furthermore, intelligent transport can be extended to some adjacent areas (parking management, station management). All these detailed intelligent management can greatly enhance the regional transport beneﬁt and decrease the operating cost for whole system.

7 Conclusion With the expansion of urban space and the intensiﬁcation of competition, regional synergetic development becomes an unavoidable trend. The characteristics of regional transport demand are changing constantly, and requirement for united coordination

28

S. Yao and L. Liu

mechanism straight climb. Accordingly, efforts to both hardware facilities as continuous investment on transport infrastructures and software construction as transport management coordination system are important. A mature regional transport network can be support of regional interconnection and cooperation. It has a certain contribution to land saving, resources optimal allocation, environmental protection and efﬁciency improvement for individuals’ daily life. Acknowledgement. This article is supported by National Key R&D Program of China (2017YFB1200702).

References 1. Zhang, Q., Chen, G., Wen, H.: Strategic research on promoting trafﬁc integration of Pearl River Delta. Sci. Technol. Manag. Res. 30(16), 65–68 (2010). (Chinese) 2. He, L., Zheng, T., Yuan, Q.: Research on the coordination mechanisms of transport projects in the Guangdong, Hong Kong and Macao Great Bay Area. China Transp. Rev. 40(4), 12–15 (2018). (Chinese) 3. Zhou, C., Deng, H., Shi, C.: A study on synergic development of Guangdong Hong-Kong Macau Greater Bay Area. Planners 34(4), 5–12 (2018). (Chinese) 4. Liu, C.: Characteristics, problems and strategies of transportation from big cities to metropolitan areas: a case study of Hangzhou. In: Urban Planning Society of China, Government of Guiyang. Inheritance and Transformation – Collection of Papers in Annual National Planning Conference 2015(05 urban transport planning), p. 12 (2015). (Chinese) 5. Deng, H.: Construction of integrated transport system in Guangdong-Hong Kong-Macau Grand Bay Area. China Ports (5), 56–59 (2017). (Chinese) 6. Fu, J., Xu, Y.: Thoughts on trafﬁc integration. J. Shandong Jiaotong Univ. 17(1), 28–32 (2009). (Chinese) 7. Hou, B., Huang, Z., Chen, X., Fan, C.: Spatial cognitive differentiation of regional cooperative development of cultural tourism. Tour. Tribune 28(2), 102–110 (2013). (Chinese) 8. Zhongshan Transportation Bureau. Research report of modern, three-dimensional, comprehensive transportation system, strengthen urbanization and network interconnection in the Bay Area. Zhongshan Transportation Bureau (2018). (Chinese) 9. Wang, P., He, G.: Transportation integration: the development direction of integrated transportation. China Transp. Rev. (10), 10–11 (2003). (Chinese) 10. Pal, D.P.: Economic integration: systemic measures in an input-output framework. Econ. Syst. Res. 19(4), 397–408 (2007)

The Risk Attitude Research on Route Choice Problem in Regional Rail Transit Based on Risk Decision Theory Yan Zhang1,2,3, Hui Zhang1,2,3(&), and Bing Wang1,2,3 1

3

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected], {hz,wb}@swjtu.cn 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China

Abstract. This paper is talking about risk attitude on route choice problem in regional rail transit on the premise that the departure time and transportation mode are determined. An experiment and a survey are designed on the basis of the certainty equivalent method. The result shows that for positive lottery travelers are risk seeking when there’s low probability of larger gain; they are risk aversion when there’s high probability of smaller gain; for negative lottery, it is a deterministic decision; for mixed lottery, the travelers have threefold risk attitudes. Keywords: Regional rail transit Attitude Lottery

Route choice Risk

1 Introduction The researches of route choice behavior are mainly based on the Deterministic Decision Theory (DDT) and Risk Decision Theory (RDT). In the following part, risk theory contains two main parts: risk and uncertainty. The typical theory of deterministic decision theory is the Wardrop Principles [1]. The Wardrop Principles assume that: • traveler all wants to maximize utility (they all want to choose the route with the smallest travel time) • The utility function has only one variable—the travel time • Each route’s travel time can be objectively known and is the function of the trafﬁc flow or passenger volume The Wardrop Principles’ assumptions make the route choice easy to operate, however, far away from the reality. To cope with this problem, the Random Utility

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 29–36, 2019. https://doi.org/10.1007/978-3-030-04582-1_4

30

Y. Zhang et al.

Theory (RDU) and RDT emerged at the right moment. RDU is the theory has both the characteristics of DDT and RDT. RDU assumes that: • travelers all want to maximize the utility. • The utility function’s variables can contain the travel time or the generalized cost. • The traveler and the observer are treated separately. Take the attributes which are known by the traveler and unknown by the observer in to consideration. And represent the attributes by random variable in the utility function. The RDT considers the attributes which are unknown by both the traveler and observer. So far, the theories which are extensively used in the transportation ﬁeld are Expected Utility Theory (EUT) and Cumulative Prospect Theory (CPT) [2, 3]. The theories of RDT all take money as the target. The differences lie in the utility expression of the alternatives, the probability of the outcomes of each alternative as well as the way to express risk attitude from utility and (or) probability. EUT assumes that the utility of alternatives is the ﬁnal wealth after the choice of the alternative, the probability of each outcome is linear. EUT describes the risk seeking, risk aversion and risk neutral by the convex utility function, concave utility function and linear utility function. CPT assumes that the utility of the alternatives are the gains and losses correspond to a certain reference point, the probability of each outcome is described by the probability weighting function which is nonlinear. The risk attitude is described by both the value function and the probability weighting function, and has fourfold pattern: risk seeking of high probability loss and small probability of gain while risk aversion of high probability of gain and small probability of loss. In this paper, we only talked about the risk, while uncertainty is not in the scope of this paper. Weber [4] has proposed that the decision maker’s risk attitude can impose great influence on their decision behavior of risk decision problem. The risk attitude is the main factor that causing individual difference. Xu [5] proves the risk attitude has great influence on the utility of traveler from the data of a survey. Dohmen [6] has proved that the individual has obvious difference when facing the risk through the study of the individual’s risk attitude distribution and its influence factor based on the panel data on socio-economic survey. Wang and Guan [7] discovered that travelers have different risk attitude under different decision scenario and the risk attitude of arriving early has relation to the individual caption. It can be seen from the existed researches: the researches focus mainly on road trafﬁc ﬁeld, partly on urban rail transit ﬁeld and hardly any on regional rail transit ﬁeld; Researches mainly apply the theories of behavioral economics in the transportation ﬁeld without analyze the different features between the alternatives of this two different research ﬁelds. Therefore, it has great practical meanings to study the traveler’s risk attitude on route choice behavior in regional rail transit based on the Risk Decision Theory.

The Risk Attitude Research on Route Choice Problem

31

2 Deﬁnition So far, the applications of Risk Decision Theory on road transportation and rail transit are all based on a basic consequential assumption: There’s no difference between the probability of each outcome brought out by the simple lottery or compound lottery; the decision maker’s concern is all on the simpliﬁed lottery on the ﬁnal wealth for any risk alternative [8]. The simpliﬁed lottery is expressed in the pattern of the simple lottery. A simple lottery is deﬁned as: L ¼ ð x 1 p1 ; x 2 p2 ; ; x n pn Þ

ð1Þ

Where pn 0, and p1 ; p2 ; P pn are the respective probabilities corresponding to the outcomes x1 ; x2 ; xn , and n pn ¼ 1. Given two simple lotteries, L1 and L2 , a lottery of lotteries, say, ðL1 ; p; L2 ; 1 pÞ, where p 2 ð0; 1Þ, is known as a compound lottery [9]. Therefore, this paper expressed the risk alternative as the simple lottery. The simple lottery was expressed as ðx; p; y; qÞ, which meant there’s p probability to cost x minutes and q probability to cost y minutes. Or it was expressed as ð xÞ to cost x minutes for 100%. EV refers to the product of each outcome and its probability of a simple lottery L when the decision maker is assumed to be in accordance with the EUT. EV ðLÞ ¼

Xn i¼1

pi x i

ð2Þ

If the simple lottery L ¼ ðx1 ; p1 ; x2 ; p2 ; ; xn ; pn Þ’s utility UðLÞ equals to the utility UðCL Þ of another simple lottery ðCL ; 1Þ which has a certain outcome, CL is the certainty equivalent of the lottery L. The decision maker is risk neutral if CL ¼ EV ðLÞ; risk aversion if CL \EV ðLÞ; risk seeking if CL [ EV ðLÞ.

3 Assumption Regional rail transit system is composed of urban rail transit subsystem, suburban railway subsystem, railway subsystem (only refers to relevant system serving passengers), intercity railway subsystem and high speed railway subsystem. This paper mainly studied the cooperative route choice behavior of urban rail transit subsystem and intercity railway subsystem, hereinafter referred to as the regional rail transit route choice behavior (RCB). The regional rail transit RCB is a kind of risk decision which is different from the risk decision of the road trafﬁc system and the risk decision of behavioral economics. In the route choice behavior research of road trafﬁc ﬁeld, the research object is the commuter. Due to the ofﬁce work features, the traveler doesn’t want to arrive at the work place as early as possible on the condition that the departure time is determined. The traveler wants to arrive at the ofﬁce in the scope of the satisfactory period. If there is any probability that the traveler will be late, he/she always thinks the less time he/she is late the better. Due to the exclusive right-of-way of regional rail transit, the traveling time in the sections (transfer time is not included) has less uncertainty. Therefore, the uncertainty is mainly caused by the transfer time inner subsystem and inter subsystems. This paper assumed that travelers did not need to

32

Y. Zhang et al.

queue up in the original station and destination station. Therefore, because of the uncertainty of transfer time and the unique feature of intercity railway waiting hall which is different from the work ofﬁce, it was assumed that the traveler behavior has the following features: • On the condition that the departure time is determined, the traveler wants to arrive at the waiting hall as early as possible. • If there’s probability that the traveler will be late, no matter how much the late time is, the traveler can not catch the designated train. Therefore, even if the late time is different, the traveler’s utility is close to a constant. This paper do not take the ticket endorsement in to consideration. • The route choice behavior of regional rail transit is different from the behavioral economics in the aspect of alternative target, preference and risk attitudes. The alternative target of the behavior economics decision behavior is money which is the more the better. However, the alternative target of the regional rail transit route choice behavior is the travel time. The target is more than 0 and has a lower bound. When the travel time is more than the time difference between the departure time of the traveler and the departure time of the intercity train, the traveler is late which is loss for the traveler, otherwise, the traveler is early which is gain for the traveler. • The preference of the traveler has different features corresponds to different properties of positive lottery, negative lottery and mixed lottery. Positive lottery means that the travel time under each probability is less than or equals to the time difference. Negative lottery is on the contrary. Mixed lottery means that the travel time under some probability is less than or equals to the time difference, while the other travel time is more than the difference time.

4 Experiment and Survey In accordance with the assumption part, we designed three different experiments to analyze the risk attitude of positive lottery, negative lottery and mixed lottery. 4.1

Positive Lottery

We used the Certainty Equivalent Method to design the experiment to verify the assumption for positive lottery. There were 58 subjects (28 ladies and 30 gentlemen). Before the experiment, we invited some researchers and teachers who had some experimental experience to join in the Certainty Equivalent Experiment and gave some useful suggestions to the experiment process and then made some improvement. Then we invited all the subjects to get into a classroom, and asked them to make choice of the test questions in the experiment after the experiment explanation. Finally, we got 58 results. We analyzed the results by SPSS. The experiment included 35 questions about the choice between 2 routes which had different travel time and corresponding probability. The problem is shown in Table 1.

The Risk Attitude Research on Route Choice Problem

33

Table 1. Choice between routes of positive lottery Question

Option

If you need to take the subway to transfer to the intercity train to do some business on another city which is too important to be cancelled. You have started at 9:00 a.m., you just need to arrive at the intercity railway station at 10:00. It means that you have 60 min to travel, which of the following route you will choose? The minimum travel time is 20 min. A. p’s probability to cost 60 min; q’s probability to cost 20 min. B. costs x with certainty

The value of ðp; qÞ expressed in route A had 7 groups of values: (99%, 1%), (90%, 10%), (75%, 25%), (50%, 50%), (25%, 75%), (10%, 90%) and (1%, 99%). x’s value expressed in route B had 5 values: 60, 50, 40, 30 and 20. This experiment had two rounds to make options in question alike in Table 1 [10]. In the ﬁrst round, the aim was to ﬁnd the changing point of option from A to B or from B to A. In the send round, divided the two nearest x points of the option changing point into 5 equal time sections and then replace x with the new time points to ﬁgure out the more precise changing point. Then we got the certainty equivalent of route A. We also showed 6 questions twice to make sure the subjects do not make choice randomly. The median of the 58 subjects’ CE and the EV calculated for each lottery based on EUT are shown in Table 2. The correlation analysis of the repeated questions showed that the average of the correlation coefﬁcients of the 6 repeated questions is 0.54. It showed that the subjects didn’t choose the option randomly. We could also tell from Table 2, the subjects were risk seeking when there was low probability of larger gain ðq 0:5Þ; they were risk aversion when there was high probability of smaller gain ðq [ 0:5Þ. The relationship between the CE and EV is shown in Fig. 1. Table 2. CE and EV of the positive lottery Lottery 0.99 0.9 0.75 0.5 0.25 0.1 0.01 0.01 0.1 0.25 0.5 0.75 0.9 0.99 CE 53 49 39 35 33 27 23 EV 59.6 56 50 40 30 24 20.4 Note: the ﬁrst probability of lottery row’s each cell is the value of p and the second probability is the value for q.

4.2

Mixed Lottery

For mixed lottery, we designed several questions for the subjects to choose shown in Table 3. The value of ðp; qÞ expressed in route A had 7 groups of values: (99%, 1%), (90%, 10%), (75%, 25%), (50%, 50%), (25%, 75%), (10%, 90%) and (1%, 99%). There were 58 interviewees (28 ladies and 30 gentlemen). The median of the 58 subjects’ CE and the EV calculated for each lottery based on EUT are shown in Table 4. The correlation analysis of the repeated questions showed that the average of

34

Y. Zhang et al.

70

expected travel Ɵme

60

59.6 56 53 49

50 40

50 39

40 35

30

CE

33 30

27 24

20

23 20.4

EV

10 0 0.01

0.1

0.25 0.5 0.75 probability

0.9

0.99

Fig. 1. CE and EV of the positive lottery Table 3. Choice between routes of mixed lottery Question

Option

If you need to take the subway to transfer to the intercity train to do some business on another city which is too important to be cancelled. You have started at 9:00 a.m., you just need to arrive at the intercity railway station at 10:00. It means that you have 60 min to travel, ﬁll in the blank of B that you think will equal to A? The minimum travel time is 20 min. A. p’s probability to cost 80 min; q’s probability to cost 20 min. B. cost with certainty

Table 4. CE and EV of mixed lottery Lottery 0.99 0.9 0.75 0.5 0.25 0.1 0.01 0.01 0.1 0.25 0.5 0.75 0.9 0.99 CE 80 80 65 50 35 20 20 EV 79.4 74 65 50 35 26 20.6 Note: the ﬁrst probability of lottery row’s each cell is the value of p and the second probability is the value for q.

the correlation coefﬁcients of the 6 repeated questions is 0.55. We could also tell from Table 4, the subjects were risk aversion when there was high probability of loss; risk seeking when there was high probability of gain; risk neutral when the probability of loss is between 0.25 and 0.75. The relationship between the CE and EV is shown in Fig. 2.

The Risk Attitude Research on Route Choice Problem

35

90 80

80

70

79.4

80 74

CE or EV

60

65

65

50

50

40 30

50 CE 35

35

20 26

20

20

EV

20.6

10 0 0.99

0.9

0.75

0.5

0.25

0.1

0.01

probability of cost 80 minutes Fig. 2. CE and EV of the mixed lottery

4.3

Negative Lottery

For the negative lottery ðx; p; y; qÞ, the outcome x and y meant that the subject could not catch the train. In the experiment, we asked the subjects to choose if they would deﬁnitely be late. All of them chose to cancel this trip. Thus the negative lottery equaled to the deterministic decision, there was no risk here. Therefore, there was not risk attitude to be studied.

5 Conclusion and Discussion According to the results of the experiment and survey, we can conclude that for the positive lottery, the travelers are risk seeking when there’s low probability of larger gain; they are risk aversion when there’s high probability of smaller gain; for negative lottery, it is a deterministic decision; for mixed lottery, the travelers were risk aversion when there was high probability of loss; risk seeking when there was high probability of gain; risk neutral when the probability of loss is between 0.25 and 0.75. This paper discussed the distribution of the risk attitude of route choice problem in the regional rail transit. It is worthy of exploring the key influence factors and their correlation. This study’s object is a single trip, the day-to-day travel is the next research step. Acknowledgements. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351, 71761023), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway

36

Y. Zhang et al.

Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK0000028-ZF, 2017-RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Shefﬁ, Y.: Urban Transportation Networks: Equilibrium Analysis with Mathematical Programming Methods. Prentice-Hall, Inc., Englewood Cliffs (1985) 2. Kahneman, D., Tversky, A.: Prospect theory: an analysis of decision under risk. Econom. J. Econom. Soc. 47, 263–291 (1979) 3. Tversky, A., Kahneman, D.: Advances in prospect theory: cumulative representation of uncertainty. J. Risk Uncertainty 5, 297–323 (1992) 4. Weber, E.U., Blais, A.R., Betz, N.E.: A domain-speciﬁc risk-attitude scale: measuring risk perceptions and risk behaviors. J. Behv. Decis. Making 15, 263–290 (2002) 5. Xu, H., Zhou, J., Chen, X.: Analysis and demonstration of the traveler’s route choice behavior rule based on the prospect theory. J. Transp. Syst. Eng. Inf. Technol. 6, 95–101 (2007) 6. Dohmen, T., Falk, A., Huffman, D.B., Sunde, U., Schupp, J., Wagner, G.G.: Individual Risk Attitudes: New Evidence from a Large, Representative, Experimentally-Validated Survey, IZA Discussion Papers 1730, Institute for the Study of Labor (IZA) (2005) 7. Wang, K., Guan, H., Yan, H.: Risk preference survey and analysis in commuter route choice behavior. J. Beijing Univ. Technol. (5), 762–767 (2016) 8. Mas-Colell, A., Whinston, M.D., Green, J.R.: Microeconomic Theory. Oxford University Press, Oxford (1995) 9. Dhami, S.: The Foundation of Behavior Economic Analysis. Oxford University Press, Oxford (2016) 10. Zeng, J.: An experimental test on cumulative prospect theory. J. Jinan Univ. (Nat. Sci.) (1), 44–47, 65 (2007)

The Topological Structure of Chengdu Metro Network Based on Complex Network Theory Feng Xue1,2 and Chuan-Lei He1(&) 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected] 2 National United Engineering Laboratory of Integrated and Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. With the rapid development of urban rail transit, it is more and more necessary to study its complex network structure. Taking Chengdu which is an important city in the west as an example, calculating its static statistic index under the complex network. The network structure has 136 nodes, 146 lines under the space L model, and the average degree is 2.147, the average path length is 13.1465, network number is 0.091, network connectivity is 0.0159, network efﬁciency is 0.1237. The overall network is a tree-like structure with small-world characteristics. The index of network number, efﬁciency and connectivity is not very important, which shows that the utilization ratio is not enough, the connection between the nodes is not open, and there is a certain waste on the topology structure, which provides the development and progress space for the planning of the future lines. Keywords: Urban rail transit Statistical index

Complex network Topology structure

1 Introduction With the rapid development of Chengdu’s economy and the continuous advancement of urbanization process, the Chengdu Metro has become a complex network, which operated six lines, the total length of the line is about 196.48 km, covering all areas of Chengdu, gradually radiating the surrounding counties and urban areas. Applying complex network theory to analyse the complexity of the Chengdu Metro has a certain guiding signiﬁcance to the real network planning. In the analysis of the topological structure of transportation network, many scholars and experts have applied complex network theory to study it, and have achieved relatively fruitful results. Sienkiewicz [1] analysed the topological structure of Poland’s 21 public transport network in a city, he showed that their degree distribution obeys the power law distribution or the exponential distribution; The road network topology of This research was supported by the National Key R&D Program of China (2017YFB1200702). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 37–47, 2019. https://doi.org/10.1007/978-3-030-04582-1_5

38

F. Xue and C.-L. He

20 large cities in Germany is statistically analyzed by Lammer [2], and it is found that the distribution of trafﬁc flow has power law; Qing [3] studied the complexity of the topology of Chongqing Metro Network, and pointed out that a small number of key nodes would cause great losses to the network. This paper researches the topological structure of Chengdu Metro Network under Space L model based on complex network theory, calculates its statistical index, With the trafﬁc operation of Chengdu Metro Line 7, the network of Chengdu Metro has been further improved. Therefore, it is necessary to apply the complex network theory in the topological structure of the ‘well + ring’ type Chengdu Metro Network.

2 Topological Properties of the CMN Based on Complex Network Theory 2.1

Construction of Urban Rail Transit Network

The station and track line are the basic elements of the metro network, and the lines are connected to each other, so the metro network can be regarded as a complex network composed of track lines and stations. The metro network topology model is mainly based on Space L model, which is a line network, the node in the deﬁnition network represents the metro station, the side represents the line between the stations. Chengdu Metro Network (CMN) refers to the urban rail transit network serving in Chengdu and its surrounding areas. The CMN has 6 lines and 136 stations based on the data in April 2018, its node numbers and topology are shown in Fig. 1. Its ‘well + ring’ urban rail network structure can be regarded as a complex network composed of track lines and stations. The topology network of CMN is constructed by under Space L model through Gephi in Fig. 2.

Fig. 1. CMN operation chart (April 2018)

Fig. 2. CMN topology network

The Topological Structure of Chengdu Metro Network

39

Through the Fig. 2, the complex network status of CMN can be clearly understood. The degree of node degree is also expressed in different colors, which shows the importance of different nodes and the connection between nodes. In this paper, the Space L method is used to construct the topology network of CMN. Based on the complex network theory, many statistical characteristics are analyzed, such as the node degree and degree distribution, the average path length, the clustering coefﬁcient, the connectivity, the network number and so on. From the quantitative point of view, the urban rail transit network in Chengdu is evaluated and analyzed. 2.2

Complex Network Theory

Based on complex networks, we calculate all kinds of static statistical indexes of CMN, ﬁrst calculate its node degree and degree distribution [4]. The calculation of the degree and the cumulative probability model are as follows: ki ¼

X

Pðk [ k 0 Þ ¼

aij

1 X k¼k0

j

pk i

ð1Þ

where aij is the element of the network adjacency matrix A, to denote the existence of edges between two points, if there exists an edge between point i and point j, then aij ¼ 1; pki is the probability of ki . The average path length [5] can be deﬁned as the number of the shortest paths between two nodes, and the diameter of the network is the maximum distance between any two points; thus, the average length L of the network is deﬁned as the average of the distance between all nodes, and the average path length is calculated as: D¼

X 1 dij NðN 1Þ i6¼j

ð2Þ

It is deﬁned here as the minimum number of nodes connecting 2 nodes. Clustering coefﬁcient also called clustering coefﬁcient, the aggregation of nodes in the network is described, which reflects the dense and intensive network. For the calculation of clustering coefﬁcient of Chengdu metro network, ﬁrst, through a quantitative way, the formula is as follows: Ci ¼

2Ei ki ðki 1Þ

ð3Þ

The idea of the algorithm is as follows: to ﬁnd all adjacent nodes of a node, these adjacent nodes form a subgraph, then draw the adjacency matrix of the subgraph from the adjacency matrix A of the network, calculate the number of the edges of the subgraph, then calculate the clustering coefﬁcient of the node according to the deﬁnition of the clustering coefﬁcient, and then seek the clustering coefﬁcient of all nodes. The average value of the clustering coefﬁcient of the whole network is calculated.

40

F. Xue and C.-L. He

The betweenness includes the betweenness of the nodes and the betweenness of the edges [6]. The betweenness of the nodes is mainly discussed here. In urban rail transit network, the betweenness of nodes represents the role and influence of nodes in the whole network. The more the shortest path through a node, the larger the betweenness of the node is, the more important it is in the network. The calculation formula is as follows: Bi ¼

X Dkj ðiÞ Dkj k6¼j2G

ð4Þ

where Bi is the betweenness of the node i, Dkj is the number of the shortest paths between nodes k and j, Dkj ðiÞ is the number of the shortest paths through node i. By calculating the clustering coefﬁcient of the Chengdu metro network, it can be seen that as a special network structure, this kind of urban rail transit network does not have the characteristics of clustering. Therefore, the connectivity in trafﬁc geography can be used to replace the clustering coefﬁcient C. The meaning of connectivity and clustering coefﬁcient are close to both the ratio of the actual side to the potential side, and the difference is that the clustering coefﬁcient is the local characteristic (only considering the neighbor nodes), and the connectivity is the global characteristic (considering the whole network). The calculation formula of connectivity is as follows [7]. c¼

eG eGmax

ð5Þ

where eG is the edge number of the network G, eGmax is the biggest edge number of the network G. In addition to connectivity, there are concepts of network standardization efﬁciency E ðGÞ and mapping efﬁciency Ec to measure the fault tolerance of a plane network, which can reflect the effect of each node on the network effectiveness and the average fault tolerance for the node failure. EðGÞ ¼ 1=N Ec ¼

X 1 1 NðN 1Þ i6¼j2G dij

P

P

E ðGi Þ

i2G

E ðG Þ

¼

ð6Þ

E ðG i Þ

i2G

N E ðG Þ

ð7Þ

Where dij is the shortest path length between node i and j, N is the node number, is the supplement efﬁciency of every node i.

The Topological Structure of Chengdu Metro Network

41

3 Results 3.1

Node Degree and Degree Distribution of the CMN

The following Table 2 is obtained by Matlab programming (Table 1). Table 1. Degree, degree distribution and cumulative distribution of CMN Degree value Degree distribution Cumulated degree distribution Average degree

1 2 3 4 0.066 0.824 0.007 0.103 0.066 0.890 0.897 1 2.147

The degree and degree distribution of CMN is shown in Figs. 3 and 4.

1

4

0.9 3.5

0.8 Cumulated degree distribution Degree distribution

0.7 Probability P(K)

Degree value

3 2.5 2 1.5

0.6 0.5 0.4 0.3

1

0.2

0.5 0

0.1 0 0

20

40

60 80 Node number

100

120

140

Fig. 3. Degree distribution diagram

1

2

3

4

Degree value

Fig. 4. Degree diagram

probability

distribution

According to the above charts, it is easy to know that the topology of CMN is common as the average degree is 2.147, the transfer convenience is not high and the network is easily paralyzed when encountering major dangerous attacks. Some of the larger values are North Railway Station, Tianfu Square and other transfer stations. 3.2

Average Path Length of the CMN

After ﬁnding the adjacency matrix between nodes, using the shortest path algorithm of Floyd, the shortest path between each node is obtained by Matlab programming, and then the shortest path distribution data is shown in Table 3. The shortest path distribution and cumulative probability distribution diagram are drawn by Matlab, as follows in Fig. 5.

42

F. Xue and C.-L. He Table 2. The shortest path distribution of CMN (part) The shortest path value 1 2 11 12 23 24 40 The average path length

Probability distribution Cumulative probability distribution 0.016 0.016 0.022 0.038 0.056 0.471 0.054 0.525 0.019 0.899 0.017 0.917 0.000013 1.000 13.1465

0.06 1 0.9 0.8

0.05 Probability P(Dij)

0.7

Probability P(Dij)

0.04

0.6 0.5 0.4 0.3 0.2

0.03

0.1 0

0

5

10

15 20 25 30 The shortest path length Dij

35

40

0.02

0.01

0

0

5

10

15 20 25 30 The shortest path length Dij

35

40

45

Fig. 5. The shortest path distribution diagram of CMN

The average path length D ¼ 13:1465, which shows that when the travel volume of each node is equal, the average trip of a trip passes about 13 intervals (stations). Compared with the size of the network node, the average path is in line with the standard of ‘small world’ and the characteristics of random network, which means the average path of the network is far smaller than the total node number of its network scale by 136. From the above table, it is easy to know that the longest distance between any two stations is 36 stations. About 52.5% of the passengers can arrive within 12 stations, and nearly 90% passengers can arrive within 23 stations. 3.3

Clustering Coefﬁcient of the CMN

It is ﬁnally found that the clustering coefﬁcient of each node is 0 by using Matlab programming, so the average clustering coefﬁcient of the CMN is 0. Another method of analysis topology can be used to verify it, it is not difﬁcult to ﬁnd out that the

The Topological Structure of Chengdu Metro Network

43

connection between network nodes is relatively sparse and does not form three or more nodes from the CMN topological diagram of Figs. 1 and 2. It isn’t connected to a ‘triangle’ or even more complex structure, so no nodes in the target node can be connected, that means the number Ei of connections between the node i and all adjacent nodes is 0, which makes the numerator of clustering coefﬁcient is 0, so the clustering coefﬁcient of all nodes is also 0. 3.4

Betweenness of the CMN

The distance of any two nodes, the average shortest path length and the shortest path number among the nodes in the CMN are obtained. The shortest path number is used to calculate the betweenness value. Matlab is used to draw the betweenness distribution of the CMN as follows in Fig. 6.

0.4 0.35

Betweenness B

0.3 0.25 0.2 0.15 0.1 0.05 0

0

20

40

60 80 Node number n

100

120

140

Fig. 6. Betweenness distribution diagram of the CMN

As can be seen from the Fig. 6, the betweenness of most nodes are between 0.05– 0.15, the top 20 nodes in the number of betweenness are arranged as shown in the following Table 4. It is worth noting that the Gao Xin Station, the City of Finance, the Hatchery, the Jincheng Square and the Century City are not transfer stations, but the value of their betweenness is relatively large, indicating that the node is more important. It is obvious that the betweenness and degree are not necessarily related, and the betweenness of some nodes with a degree of 2 is also very high, which can be used as a key or hub node for protection.

44

F. Xue and C.-L. He

Table 3. The number of twenty nodes betweenness in the front of the Chengdu metro network Node number 13 (South Railway Station) 95 (Cultural Palace) 14 (Gao Xin Station) 15 (The City of Finance) 16 (Hatchery)

Betweenness Degree Node number 0.358 4 6 (mule and Ma City)

17 (Jincheng Square) 18 (Century City) 81 (Taiping garden) 56 (Chengdu East passenger station) 19 (Tianfu three street)

3.5

Betweenness Degree 0.209 4

0.271

4

20 (Tianfu ﬁve street)

0.199

2

0.265

2

21 (Hua Fu Road)

0.187

2

0.254

2

22 (Four River)

0.178

3

0.244

2

0.173

4

0.233

2

47 (Hospital of large province of traditional Chinese Medicine) 103 (Huai tree shop)

0.170

4

0.222

2

44 (A whole world)

0.169

4

0.221

4

10 (Provincial Gymnasium)

0.167

4

0.217

4

7 (Tianfu Square)

0.166

4

0.211

2

126 (The fairy tree)

0.164

2

Reliability Index of the CMN

The connectivity of Chengdu metro network is calculated by Matlab programming. c¼

eG 146 ¼ 0:0159 ¼ eGmax ð136 135Þ=2

ð8Þ

The standardization efﬁciency of the network. EðGÞ ¼

X 1 1 ¼ 0:1237 NðN 1Þ i6¼j2G dij

ð9Þ

These two indexes of network efﬁciency and connectivity are used to measure the reliability or the robustness of the CMN after network failure. The index of CMN is not large. Finally, the reliability of Chengdu rail transit network can be analyzed according to the changes of these indicators. The calculation efﬁciency EðGi Þ for each node caused by the damage of different nodes, and the connectivity degree c of the network as a whole after the damage of each node are changed, respectively (Figs. 7 and 8).

The Topological Structure of Chengdu Metro Network 0.125

45

0.0158

0.0157

0.0157 Network connectivity

Supplement efficiency E(Gi)

0.12

0.115

0.11

0.0156

0.0156

0.0155 0.105

0.0155

0.1

0

20

40

60 80 Node number n

100

120

140

0.0154

0

20

40

60 80 Node number n

100

120

140

Fig. 7. Distribution efﬁciency of each node in Fig. 8. Change diagram of network connectivChengdu Metro Network ity after the damage of Chengdu metro network nodes

The degree of variation of network efﬁciency and connectivity after the failure of network nodes is given. Take the ten nodes with the largest change as an example in Tables 5 and 6. Table 4. Network connectivity changes after failure of network nodes in Chengdu Metro (top 10) Node

Connectivity

Degree

Betweenness

0.0155 0.0155 0.0155 0.0155 0.0155 0.0155 0.0155

Change rate 2.71% 2.71% 2.71% 2.71% 2.71% 2.71% 2.71%

3 (North Railway Station) 6 (mule and Ma City) 7 (Tianfu Square) 10 (Provincial Gymnasium) 13 (South Railway Station) 44 (a whole world) 47 (Hospital of large province of traditional Chinese Medicine) 50 (Chunxi Road) 56 (Chengdu East passenger station) 71 (SIMA bridge)

4 4 4 4 4 4 4

0.123 0.209 0.166 0.167 0.358 0.169 0.173

0.0155 0.0155 0.0155

2.71% 2.71% 2.71%

4 4 4

0.159 0.217 0.119

After the failure of the node, the change of connectivity is small, and the size of the connectivity and the rate of change are closely related to the degree of the degree. The nodes with the same degree are also the same in the network connectivity, but the network efﬁciency and the rate of change are not necessarily related to the degree of the nodes, although most of the network efﬁciency changes. Some of the nodes with higher degree of transformation are all the largest transfer points, but such as node 14

46

F. Xue and C.-L. He

Table 5. Network efﬁciency changes after damage failure of network nodes in Chengdu Metro (top 10) Node 13 (South Railway Station) 14 (Gao Xin Station) 95 (Cultural Palace) 15 (the city of Finance) 16 (hatchery) 56 (Chengdu East passenger station) 17 (Jincheng Square) 18 (Century City) 94 (Qingjiang West Road) 44 (a whole world)

Network efﬁciency 0.1029 0.1065 0.1069 0.1077 0.1088 0.1095

Change rate 16.84% 13.90% 13.60% 12.90% 12.04% 11.50%

Degree

Betweenness

4 2 4 2 2 4

0.358 0.265 0.271 0.254 0.244 0.217

0.1098 0.1106 0.1108 0.1111

11.27% 10.56% 10.41% 10.17%

2 2 2 4

0.233 0.222 0.163 0.169

Table 6. The characteristic index value of the CMN Index The number of nodes The number of lines Average degree Average path length Average clustering coefﬁcient Average betweenness Connectivity Network efﬁciency

Value 136 146 2.147 13.1465 0 0.091 0.0159 0.1237

(Gao Xinzhan), the change of network efﬁciency is 13.90%, the degree is only 2, and it is a general intermediate node. This should also be regarded as a key node to strengthen protection and repair, ensure smooth flow and achieve quick rescue.

4 Summary of Static Characteristics of the CMN The following characteristics of the CMN are separately listed in Table 5. As can be seen from the above Table 5, there are 136 nodes in CMN, which are connected by 146 sides. The average path length is 13.1465. It conforms to the ‘small world’ network standard, and the clustering coefﬁcient is 0. It shows that the connection of adjacent nodes is poor and the connection between each other is less, and the network of Chengdu Metro presents a tree structure. Some index such as efﬁciency and connectivity are small, which reflects the lack of network topology, and the next step of planning needs to be strengthened.

The Topological Structure of Chengdu Metro Network

47

5 Conclusions Through the establishment of the above model and the calculation of static statistical index, the following conclusions can be obtained for the application of complex network in the CMN. (1) Based on the complex network theory, the topological characteristics of the CMN are analyzed from static statistical indexes such as node degree, average path length, clustering coefﬁcient, and betweenness. The index calculation results show that the network sites of CMN are less connected and have a certain waste in the topology structure. (2) Through the analysis of network efﬁciency and connectivity of the CMN, it can be found that not only do we need to protect the transfer stations such as the South Railway Station (13), but also we need to strengthen some intermediate nodes but high-betweenness such as Gao Xin Station (14) to maintain the key nodes. (3) This paper is limited to the Chengdu Metro Network, it can be combined with other trafﬁc networks, such as Chengdu Road, Railway Network and other comprehensive analysis. A quantitative index is calculated for the friendly degree of Chengdu trafﬁc network construction. We need accelerate the improvement of the comprehensive transport network system framework, and promote the coordinated development of urban areas. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Sienkiewicz, J., Holyst, J.A.: Public transport systems in Poland: from Bialystok to Zielona Gora by bus and tram using universal statistics of complex networks. Acta Phys. Polonica 36 (5), 310–317 (2005) 2. Lämmer, S., Gehlsen, B., Helbing, D.: Scaling laws in the spatial structure of urban road networks. Phys. A Stat. Mech. Appl. 363(1), 89–95 (2006) 3. Qing, Y.: Analysis of vulnerability analysis of rail trafﬁc network based on complex network theory. Chin. J. Secur. Sci. 22(2), 12 (2012) 4. Wu, J.J.: Studies on the Complexity of Topology Structure in the Urban Trafﬁc Network. Beijing Jiaotong University (2008) 5. Zhang, J., Liang, Q.H., He, X.T.: Study on the complexity of Beijing metro network. J. Beijing Jiaotong Univ. 37(6), 78–84 (2013) 6. Yin, Y., Liu, J.: The survivability of railway network. Integr. Transp. 39(4), 55–59 (2017) 7. Yu, B., Feng, C., Zhu, Q., et al.: Vulnerability analysis of China’s high speed railway network. China Saf. Sci. J. 27(9), 110–115 (2017)

Key Node Identiﬁcation Method of Chengdu Metro Network Based on Comprehensive Assessment Feng Xue1,2, Chuan-Lei He1(&), Zong-Sheng Sun1, and Xiao Yu1 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected] 2 National United Engineering Laboratory of Integrated and Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. The identiﬁcation of key nodes is conducive to improving the accuracy of network performance analysis. The topology of the Chengdu Metro Network (CMN) is analyzed based on complex network theory, a node importance evaluation index system including regional prosperity and node accessibility is constructed. Multi-comprehensive evaluation methods such as Relational Analysis, TOPSIS, Principal Components Analysis, Entropy Weight Method, etc., were used to rank 136 nodes. It is identiﬁed that 30 key nodes headed by East Chengdu Railway Station (56). At the same time, some starting, ending and intermediate stations with low degree or betweenness are equally important in urban rail transit network. Key precautions and maintenance are needed to ensure the normal operation of the CMN. Keywords: Urban rail transit Complex network

Key node Comprehensive assessment

1 Introduction The rapid network formation of Chengdu Metro makes its network structure have certain complexity, on the basis of analyzing its topology based on complex network theory, the importance of each node is quantitatively calculated to identify the key nodes and protect the key nodes, so as to ensure the safety of the operation of Chengdu Metro Network. In the key node of the trafﬁc network recognition ﬁelds, many scholars and experts applyed a variety of comprehensive evaluation method for transport network node importance index analysis, and some achievements have been made in the identiﬁcation and selection of key nodes. Qing [1] quantitatively calculated the vulnerability of each node of Chongqing Metro Network under malicious attacks, so as to identify key

This research was supported by the National Key R&D Program of China (2017YFB1200702). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 48–58, 2019. https://doi.org/10.1007/978-3-030-04582-1_6

Key Node Identiﬁcation Method of Chengdu Metro Network

49

nodes with the size of influence on network efﬁciency. Liu [2] et al. constructed the selection index of Highway Network node importance, and used the Grey Relational Analysis method to calculate the importance value of each node. Yang et al. [3] realized that the importance of Beijing Subway System nodes depend on two indicators: degree value and interface number. Yu et al. [4] improved on the basis of Yang Yuhao and solved the comprehensive evaluation value of site importance under the combination of different degree and betweenness parameters. The existing research is simple to identify and select key nodes, and there is no uniﬁed indicator system. Moreover, the evaluation method of node importance is relatively simple. In view of this, based on the analysis of Chengdu Metro Network topology, focusing on improving the identiﬁcation and selection methods of key nodes, constructing the index system of node importance. Multiple comprehensive evaluation methods, such as Grey Relational Analysis, TOPSIS, Principal Components Analysis, Entropy Weight Method, etc., are used to evaluate the joint importance. In order to make the Chengdu Metro Network performance change simulation under the malicious attacks of key nodes.

2 Chengdu Metro Network Topology Structure Chengdu Metro Network (CMN) is the urban rail transit network serving Chengdu and its surrounding areas, many famous stations such as Tianfu Square and Sichuan Gymnasium have become logo of Chengdu. The metro brings different areas of the city closer together, which greatly satisﬁes the needs of passengers for commuting, studying and traveling. Based on the data of CMN in April 2018, a total of 6 lines (lines 1, 2, 3, 4, 7, 10) were opened, the total length of the line was about 196 km, 136 stations were put into operation, the lines were divided into the front line, auxiliary line and garage line. Each station number of CMN is shown in the following Table 1.

Table 1. Station information of the CMN (part) ID Station 1 Weijianian 2 Shengxian Lake 3

4 5 6

Degree ID Station 1 46 Baiguolin 2 47 Sichuan Provincial People’s Hospital 4 48 Tonghuimen

North Railway Station Renmin Rd. 2 North Wenshu 2 Monastery Luomashi 4

Degree ID 2 91 4 92

Station Caiqiao Zhongba

Degree 2 2

2

93

West Railway Station

2

49 People’s park

2

94

2

50 Chunxi Road

4

95

West Qingjiang Road Cultural Palace

51 Dongmen Bridge 2

96

4

SWUFE University 2 (continued)

50

F. Xue et al. Table 1. (continued)

ID Station 7 Tianfu Square 8 Jinjiang Hotel 9 Huaxiba

Degree ID Station 4 52 Niuwangmiao

Degree ID 2 97

Degree 2

2

53 Niushikou

2

2

2

54 Dongda Road

2

10 Sichuan 4 Gymnasium 11 Nijiaqiao 2

55 Tazishan Park

2

Station North Caotang Road 98 Wide and Narrow Alley 99 South Taisheng Road 100 Yushuang Road

12 Tongzilin 13 South Railway Station 14 Hi-tech Zone … … 43 Yangxi Flyover 44 Yipintianxia 45 East Shuhan Road

2 2

101 Shuangqiao Road

2

2 4

56 East Chengdu 4 Railway Station 57 Chengyu Flyover 2 58 Huiwangling 2

102 Wannianchang 103 Huaishudian

2 4

2

59 Honghe

104 Lailong

2

… 2

… … … 88 Fenghuang street 2

… 2

4 2

89 Machangba 2 2 90 Intangible Cultural Heritage Park

… … 134 Southwest Jiaotong University 135 Jiulidi 136 West 2nd Road of North Railway Station

2

2 2

Through the Fig. 2, the complex network status of CMN can be clearly understood. The degree of node degree is also expressed in different colors, which shows the importance of different nodes and the connection between nodes. The nodes in the above map are numbered in Arabia, with a total of 136 nodes. The adjacency matrix of the connection between them is shown in Table 1. The nodes in the above map are numbered in Arabia, with a total of 136 nodes. The adjacency matrix of the connection between them is shown in Table 2. The metro network can be regarded as a complex network composed of rail lines and stations [1], the topology model of metro network is mainly based on Space-L and Space-P. The former is the line network, which deﬁnes nodes to represent metro stations in the network and deﬁnes sides to represent lines between nodes. The latter deﬁnes that if two nodes are on the same line, which means there is an edge between two nodes. Taking the intersection of the CMN line 1 and line 2 as an example, the model is illustrated in Fig. 1 and 2. Based on complex network theory and Space-L model, static characteristic indexes of the CMN topology are calculated, as shown in Table 3. The above graph clearly reflects the topological characteristics of the CMN. the importance of task node is related to the degree and the number of interfaces. But

Key Node Identiﬁcation Method of Chengdu Metro Network

51

Table 2. Network adjacency matrix of Chengdu Metro (part) 1 2 3 4 5 6 7 134 135 136

Sichuan Provincial People's Hospital Tonghuimen

1 0 1 0 0 0 0 0 0 0 0

2 1 0 1 0 0 0 0 0 0 0

3 0 1 0 1 0 0 0 0 0 1

4 0 0 1 0 1 0 0 0 0 0

5 0 0 0 1 0 1 0 0 0 0

134 0 0 0 0 0 0 0 0 1 0

135 0 0 0 0 0 0 0 1 0 1

136 0 0 1 0 0 0 0 0 1 0

Luomashi

People's park

Luomashi

Tonghuimen

Tianfu Square

Wenshu Monastery

Jinjiang Hotel

Line 1 Line 2

7 0 0 0 0 0 1 0 0 0 0

Wenshu Monastery

Tianfu Square People's park

6 0 0 0 0 1 0 1 0 0 0

Huaxiba

Chunxi Road

Sichuan Provincial Chunxi Jinjiang Hotel People's Hospital Road

Fig. 1. Space-L model

Huaxiba

Fig. 2. Space-P model

Table 3. The characteristic index value of the CMN Index Value Average clustering coefﬁcient 0

The number of nodes 136 Betweenness

The number of lines 146 Connectivity

0.091

0.0159

Average degree 2.147 Network efﬁciency 0.1237

Average path length 13.1465 Supplement efﬁciency 0.1183

regardless of topological structure indicators such as degree of dependence and betweenness, or vulnerability indicators like connectivity and network efﬁciency, neither can effectively identify the key nodes objectively and comprehensively due to the few references. Therefore, it is necessary to build an evaluation system of node importance, the comprehensive evaluation method is adopted to evaluate the signiﬁcance to identify the key nodes. For the key node identiﬁcation technology, the static indicators of each node can be obtained according to statistics, the importance of each node is calculated, the critical

52

F. Xue et al.

degree is evaluated and graded. Next, the evaluation system of various indicators of 136 nodes is constructed, and then comprehensive evaluation and classiﬁcation are carried out.

3 Construction of Node Importance Index System Major accidents may occur in subway network operation due to equipment failure, man-made damage, force majeure and other reasons. When these unexpected incidents occur, some road sections of the subway network will be damaged and cannot pass, ensuring the smooth operation of key stations and road sections will help maintain the orderly transportation of the entire subway network. The traditional key node identiﬁcation method generally starts from the topology of the network. After the node is damaged, the overall connectivity and efﬁciency of the network are observed. The combination of the degree and the median is analyzed. The operation is simple and feasible. In order to establish a complete node importance evaluation index system, the selected indicators need to be scientiﬁc, representative and comprehensive. To a certain extent, it not only reflects the importance of the topological structure of the nodes in the network, but also links the actual line network of urban rail transit network, manifesting the subway station distributing passengers and other key ways of connection. The selected indicators are shown in Table 4.

Table 4. Index system of node importance in the CMN First-grade Index B1 Topological structure Node importance index system

B2 Vulnerability B3 Passenger flow and equipment B4 Opportunity index

Second-grade Index C1 Degrees C2 Betweenness C3 Supplement efficiency C4 Connectivity C5 Number of passengers entering the station C6 Number of exit of station C7 The surrounding resources C8 Degree of prosperity

Topological structure and vulnerability indicators can be calculated by complex network theory formulas. Although the connectivity is an indicator of the entire network, the importance of a node can also be measured by the change of the indicator of the network after the node is damaged; Passenger flow and equipment indicators need to collect passenger flow information data from Chengdu Metro related stations. The daily arrival number of passenger station (persons/times) is indicated. The opportunity index reflects the accessibility of a certain station in passenger travel [5]. It includes two indicators of the surrounding resources and prosperity. The surrounding resources

Key Node Identiﬁcation Method of Chengdu Metro Network

53

are based on the evaluation of the hundred points based on the resources of schools, hospitals, commercial streets and tourist attractions around the station, and the prosperity is the GDP (ten thousand/person) statistics of the per capita area of the station. the data of some indicators are summarized as follows due to the limited space (Table 5).

Table 5. Index data summary Station serial number 1 2 3 4 5 6 7 8 9 10 … 136

Station name

C1 C2

C3

C4

C5

C6 C7 C8

Weijianian Shengxian Lake North Railway Station Renmin Rd. North Wenshu Monastery Luomashi Tianfu Square Jinjiang Hotel Huaxiba Sichuan Gymnasium … West 2nd Road of North Railway Station

1 2 4 2 2 4 4 2 2 4 … 2

0.1220 0.1202 0.1158 0.1204 0.1201 0.1184 0.1185 0.1206 0.1208 0.1172 … 0.1204

0.0158 0.0157 0.0155 0.0157 0.0157 0.0155 0.0155 0.0157 0.0157 0.0155 … 0.0157

7955 12453 40036 24654 21740 37099 47518 13953 32377 32342 … 2083

5 2 3 2 6 2 10 4 3 3 … 2

0 0.015 0.123 0.076 0.085 0.209 0.166 0.087 0.082 0.167 … 0.053

47 41 89 53 76 93 98 67 70 90 … 42

8.84 9.04 8.84 8.84 8.84 8.84 12.76 13.5 7.07 7.07 … 8.84

In order to complete the identiﬁcation of key nodes, correlation degree analysis among indicators is carried out.

4 Key Node Identiﬁcation Based on Combination Evaluation Method 4.1

Solution of Grey Relational Degree Analysis Method

Gray Relational Analysis (GRA) can be used to evaluate the multi index synthetically as the similar or different degree of the development trend between factors. It has the characteristics of simple operation and reliable results. The grey relational analysis method combines the quantitative and qualitative methods organically, and the calculation is convenient, and to a certain extent, the subjective arbitrariness of the decision-makers is excluded. The conclusions are more objective and have some reference value.

54

F. Xue et al.

(1) Identify comparison objects (evaluation objects) and reference series (evaluation criteria). There are m evaluation objects and n evaluation indicators, then the reference series is x0 ¼ fx0 ðkÞjk ¼ 1; 2; . . .; ng, Comparison sequence is xi ¼ fxi ðkÞjk ¼ 1; 2; . . .; ng; i ¼ 1; 2; . . .; m. (2) Determine the weight of each index value. In this paper, the corresponding weights are given by the method of expert investigation w ¼ ½w1 ; . . .; wn . (3) Calculate the grey correlation coefﬁcient: ki ðkÞ ¼

min minfx0 ðtÞ xs ðtÞg þ q max maxfx0 ðtÞ xs ðtÞg s

t

s

t

½x0 ðkÞ xs ðkÞ þ q max maxfx0 ðtÞ xs ðtÞg s

ð1Þ

t

The meaning is to compare the correlation coefﬁcient of the series xi to x0 on the kth indicator. (4) Calculate the gray weighted association: ri ¼

n X

wi ki ðkÞ

ð2Þ

k¼1

Its meaning is the grey weighted correlation degree of the ith evaluation object to the ideal object. (5) Evaluation and analysis Establishing the correlation order of the evaluation object according to the size of the gray weighted association. The greater the degree of association, the better its evaluation results. Thus, the importance of the 136 nodes is sorted. We use Resolution coefﬁcient q ¼ 0:5 to calculate the correlation coefﬁcient ki ðkÞ and grey weighted correlation degree ri . The results obtained are shown in Table 6 by using Matlab. Table 6. Correlation coefﬁcient and relevance value (part) Node 1 2 3 4 5 … 136

C1 0.333 0.429 1.000 0.429 0.429 … 0.429

C2 0.333 0.343 0.432 0.388 0.396 … 0.370

C3 0.926 0.793 0.587 0.806 0.787 … 0.806

C4 1.000 0.600 0.333 0.600 0.600 … 0.600

C5 0.351 0.363 0.455 0.399 0.389 … 0.337

C6 0.474 0.360 0.391 0.360 0.529 … 0.360

C7 0.393 0.471 0.516 0.414 0.405 … 0.372

C8 0.333 0.429 1.000 0.429 0.429 … 0.419

ri 0.548 0.473 0.517 0.486 0.507 … 0.474

Key Node Identiﬁcation Method of Chengdu Metro Network

4.2

55

Solution of TOPSIS

TOPSIS is an effective multi-index evaluation method, the method constructs positive and negative (the worst) ideal solutions for evaluation problems. By calculating the relative closeness of each scheme to the ideal scheme to sort the solution and select the optimal solution. (1) Obtain the canonical decision matrix by vector programming. (2) Construct the weighted canonical matrix. (3) Calculate positive and negative ideal solutions. (4) Calculate the distance from each solution to positive and negative ideal solutions. (5) Calculate the comprehensive evaluation index of each program. (6) Rank the pros and cons of a plan from large to small. In the established TOPSIS model, we standardize and normalize the original data, set weight vector according to expert investigation method and relevant experience, as follows. w ¼ ½0:15; 0:15; 0:15; 0:15; 0:1; 0:1; 0:1; 0:1 The positive and negative ideal solutions of 136 nodes for 8 indicators were calculated by MATLAB (Table 7). Table 7. Positive, negative ideal solution distance and comprehensive index values (part) Node 1 2 3 4 5 … 136

4.3

Positive ideal solution distance 0.069863 0.068546 0.048387 0.059858 0.056576 … 0.070634

Negative ideal solution distance 0.013519 0.010167 0.032185 0.017756 0.022825 … 0.009476

Comprehensive index values 0.162133 0.129164 0.399456 0.228773 0.287461 … 0.118292

Solution of Principal Components Analysis

Principal component analysis (PCA) is also called principal component analysis, which aims to transform the multi index into a few comprehensive indexes (i.e. principal components) by using the idea of dimensionality reduction, in which each principal component can reflect most of the information of the original variable, and the information contained is not repeated. This method introduces multiple variables into several principal components and simpliﬁcation, and obtains more scientiﬁc and effective data information at the same time.

56

F. Xue et al.

The evaluation values of 136 sections calculated by principal component analysis are as follows (Table 8).

Table 8. Principal component analysis results (part) Node 1 4 7 10 13 … 134

4.4

Evaluation value Node −1.219 2 −0.467 5 3.050 8 1.677 11 2.807 14 … … −0.729 135

Evaluation value Node −0.914 3 0.419 6 0.040 9 −0.041 12 1.377 15 … … −0.513 136

Evaluation value 1.837 1.847 0.103 0.541 1.134 … −1.022

Solution of Entropy Weight Method

The determination of index weight is mainly divided into two kinds, one is subjective empowerment, the other is objective weighting. Because subjective empowerment is subjective and arbitrary, and it is easily limited to the completeness of expert knowledge and experience, objective empowerment is based on the objective relation between data to determine weight. Entropy weight method (EWM) is the most widely used method to determine the weight objectively [6]. In order to reflect the differences between stations more intuitively, we can adopt the combination of subjective judgment and objective empowerment, grey relational model, TOPSIS model and PCA model to solve the results, and use entropy weight method to give three comprehensive evaluation methods, and get comprehensive evaluation ranking. After the calculation, the entropy weight method is used to determine the weight of the above three comprehensive evaluation methods as follows: 0.5055, 0.3145, 0.18. To sum up, the importance of 136 nodes was sorted. The key nodes are identiﬁed by taking the evaluation results of each node as the standard to measure its importance. The top 30 nodes of importance are selected as the key nodes, and their scores and indicators are shown in Table 9. It is worth mentioning that the comprehensive evaluation results are in good agreement with the complex network topology indicators. Transfer nodes with large degrees and large betweenness are generally more critical and important, and need to be protected and maintained. However, intermediate nodes such as Longquanyi and Xipu, and the intermediate nodes such as the 3rd Tianfu Street and Century City cannot be ignored. They effectively complete the agglomeration and evacuation of urban rail transit passenger flow, and the role in the urban rail network can not be neglected.

Key Node Identiﬁcation Method of Chengdu Metro Network

57

Table 9. Ranking of each node’s importance Order Node Station 1 56 East Chengdu Railway Station 2 50 Chunxi Road 3 7 Tianfu Square 4 95 Cultural Palace 5 13 South Railway Station 6 103 Huaishudian 7 75 Chengdu Second People’s Hospital 8 16 Longquanyi 9 81 Taipingyuan 10 47 Sichuan Provincial People’s Hospital 11 6 Luomashi 12 36 Xipu 13 3 North Railway Station 14 10 Sichuan Gymnasium 15 44 Yipintianxia 16 19 3rd Tianfu Street 17 18 Century City 18 14 Hi-tech Zone 19 71 Simaqiao 20 15 Financial City 21 16 Incubation Rark 22 22 Sihe 23 20 5th Tianfu Street 24 92 Zhongba 25 128 Wuhou Avenue 26 17 Jincheng Plaza 27 123 Sichuan Normal University 28 21 Huafu Avenue 29 94 West Qingjiang Road 30 12 Tongzilin

GRA TOPSIS PCA EWM Degree Betweenness 0.600 0.736 3.703 1.202 4 0.217 0.621 0.662 0.585 0.560 0.562 0.584

0.702 0.560 0.480 0.527 0.410 0.416

4 4 4 4 4 4

0.159 0.166 0.271 0.358 0.170 0.154

0.553 0.412 0.522 0.409 0.531 0.429

2.136 0.794 1 2.033 0.759 4 1.933 0.751 4

0 0.221 0.173

0.542 0.577 0.517 0.520 0.501 0.479 0.473 0.470 0.506 0.465 0.460 0.455 0.464 0.492 0.507 0.457 0.546

1.847 1.621 1.837 1.677 1.732 1.523 1.382 1.377 1.311 1.134 1.007 1.069 0.917 0.887 0.807 0.793 0.670

4 1 4 4 4 2 2 2 4 2 2 3 2 2 2 2 2

0.209 0 0.123 0.167 0.169 0.211 0.222 0.265 0.119 0.254 0.244 0.178 0.199 0.138 0.157 0.233 0.126

0.735 0.475 2 0.731 0.475 2 0.541 0.448 2

0.187 0.163 0.119

0.437 0.455 0.399 0.393 0.352 0.506 0.468 0.454 0.295 0.410 0.403 0.348 0.379 0.344 0.349 0.369 0.297

0.458 0.355 0.486 0.312 0.492 0.324

3.422 3.050 2.827 2.807 2.292 2.218

1.151 1.059 0.956 0.954 0.826 0.825

0.744 0.727 0.717 0.688 0.676 0.675 0.635 0.628 0.585 0.568 0.541 0.532 0.519 0.517 0.511 0.490 0.490

5 Conclusions Based on Grey Relational Analysis, TOPSIS, Principal Components Analysis, Entropy Weight Method etc., comprehensive assessment methods, combined with the application of complex network theory in the CMN, we can get the following conclusions and prospects:

58

F. Xue et al.

(1) The node importance evaluation index is established based on the key node identiﬁcation method, and the importance degree of 136 nodes is comprehensively evaluated, the importance degree is sorted. It is found that the initial and ﬁnal stations of Xipu and Longquanyi, degree and betweenness are not high, but can still be used as a key node to focus on prevention due to their higher passenger flow, outbound, station outbound equipment and regional prosperity. (2) The identiﬁcation of key nodes is beneﬁcial to improve the accuracy of network performance analysis. It can accurately process the performance analysis of the later network, and observe the decline trend of the performance indicators of the network by deliberately attacking the key nodes to illustrate the importance of prevention. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Qing, Y.E.: Vulnerability analysis of rail transit based on complex network theory. China Saf. Sci. J. 22(2), 122–126 (2012) 2. Liu, W., Zhu, S.Y., Zhang, J.X.: An objective method to calculate the importance of highway node. J. Chongqing Jiaotong Univ. (Nat. Sci.) 23(1), 42–44 (2004) 3. Yang, Y., Liu, Y., Zhou, M., et al.: Robustness assessment of urban rail transit based on complex network theory: a case study of the Beijing Subway. Saf. Sci. 79, 149–162 (2015) 4. Bao, Y.U., Chun, F.E.N.G., Qian, Z.H.U., et al.: Vulnerability analysis of China’s high speed railway network. China Saf. Sci. J. 27(9), 110–115 (2017) 5. Zhao, D.: Evaluation on multimodal public transit resources integration. Southeast University (2016) 6. Peng, J.X.: Modeling and empirical analysis of bus and subway weighted composite network in Nanjing. Nanjing University of Posts and Telecommunications (2017)

Research on Freights Organization Strategy Based on China-Laos Railway Yihan Wang1, Lan Liu1,2(&), and Hao Huang1 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. The completion of the China-Laos railway will enable the China and Laos to develop faster and better, and will also play a major role in promoting ASEAN economic integration. Therefore, based on the analysis of the current economic situation and social development of both China and Laos, this paper analyzes the main trade of freights between China and Laos together with the freights flow attracted by the China-Laos railway. Also this paper combines the experience and methods of domestic railway freights transport with freights organization to propose a viable freights flow. What’s more, this paper will explore the applicability of methods to provide a reference for the development of China-Laos railway international multi-modal transport. Keywords: Railway freights transport Freights organization

International multi-modal transport

1 Introduction The trade volume between China and Laos will increase substantially nowadays. The reasonable and efﬁcient freights organization can not only shorten the time of freights transportation but also improve the efﬁciency of transportation, which can better meet the needs of shippers. Therefore, a reasonable and efﬁcient freights organization is an important part of improving the competitiveness of the railway market. China’s railway freights organization model is relatively simple, and due to the railway itself, many marketing centers are passive. Eva Savesberg [1] analysis freights organization in Europe and gets the conclusion that efﬁcient freights organization depend on the multiple ways of organization. For the freights organization of scattered freights, Wang Zhimei [2] established an integer programming model to optimize the train operation based on the uncertainty of scattered freights. For the freights organization of container trains, Li Chongyu [3] analyzed the status quo of China’s container trains, and proposed the corresponding organization method according to the target This research was supported by the National Key R&D Program of China (2017YFB1200702). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 59–66, 2019. https://doi.org/10.1007/978-3-030-04582-1_7

60

Y. Wang et al.

customers. The Chi Cheng [4] is based on the development of the source of freights as the entry point to propose the freights organization optimization strategy. Zhang Yuzhao [5] discussed the freights organization strategy under the freights reform. The above research provides a freights theoretical basis for the development of China’s freights organization, but it can also be seen that the current research targets of freights organization are mostly domestic railways, but there is little involvement in the freights organization of international multimodal railways. This paper combines the experience of domestic freights transportation, from the perspective of improving the efﬁciency of railway freights transportation and meeting customer needs and expanding the transportation market, to propose the strategy and method of freights organization of China-Laos railway.

2 Major Trade Freights Between China and Laos 2.1

Trade Situation

In the past 20 years, the average annual growth rate of bilateral trade between China and Laos has reached 135.526 million US dollars. The bilateral trade cooperation and communications between China and Laos have become increasingly close together, and the total trade growth has been obvious. According to statistics from the Asian Development Bank (ADB), bilateral trade between China and Laos totaled US$24 million in 1996. Among them, China exported 23.2 million US dollars to Laos, ranking fourth, following Thailand, Japan and Vietnam; China imported 8 million US dollars from Laos, ranking second, just after Thailand. By the end of 2015, the bilateral trade volume between China and Laos had reached US$2,790.3 million, accounting for 0.6% of the total bilateral trade between China and the 10 ASEAN countries. It is China’s ninth largest trading partner in ASEAN. Among them, China’s exports to Laos were 1.234 billion US dollars, down 33.0% from 2014; China’s imports from Laos were 1.556 billion US dollars, down 12.2% from 2014. Table 1 shows the development of bilateral trade between China and Laos from 1995 to 2015.

Table 1. State of trade between China and Laos from 1995 to 2015 Years 1996 1997 1998 1999 2000 2001 2002 2003 2004

Volume of trade (million yuan) 24.0 5.2 26.8 33.1 43.7 66.7 68.5 118.3 122.1

China exports to Laos (million yuan) 23.2 4.9 19.6 24.4 37.9 59.9 59.7 108.1 110.6

China imports from Laos (million yuan) 0.8 0.3 7.2 8.7 5.8 6.8 8.8 10.2 11.5 (continued)

Research on Freights Organization Strategy

61

Table 1. (continued) Years 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

2.2

Volume of trade (million yuan) 139.1 230.7 272.5 431.4 719.9 1035.0 1248.3 1725.4 2732.6 3617.3 2790.3

China exports to Laos (million yuan) 115.9 185.6 195.2 295.5 413.9 524.1 519.3 934.5 1225.7 1839.4 1234.5

China imports from Laos (million yuan) 23.2 45.1 77.3 135.9 306.0 510.9 729.0 790.9 1547.3 1777.8 1555.8

Commodity Structure

For a long time, the commodities imported and exported between China and Laos have been dominated by raw materials and primary products. According to the United Nations Commodity Trade Statistics in 2015, the top ﬁve commodities exported by Laos to China were wood products, mineral products, copper and its products, fertilizers, rubber and their products. The number of commodities was 480 million US dollars, US$386 million, US$152 million, US$94 million and US$71 million respectively, accouting for 90.4% of China’s total export trade; China’s top ﬁve commodities export to Laos are electronics, machinery, spacecraft and their parts, vehicles and their parts, Steel products, the value of freights were 312 million US dollars, 246 million US dollars, 151 million US dollars, 123 million US dollars, 109 million US dollars respectively, accounting for 77.6% of China’s total exports to Laos. Tables 2 and 3 show the top 10 commodities exported by Laos to China in 2015 and the top 5 commodities exported by China to Laos [6]. Table 2. Top ten commodities of Laos exports to China in 2015 Catagories Minerals Woodwork Copper and its products Rubber and its products Vegetable oil, seed, fruit Grain and its products Fossil fuel Furniture, lighting, house decoration materials Malt, starch, wheat and processing products Resin and its products

Proportion (%) 43.10 31.76 10.33 8.28 2.80 1.54 0.81 0.31 0.25 0.21

62

Y. Wang et al. Table 3. Top ﬁve commodities of China exports to Laos in 2015 Catagories Electronics Mechanics Spacecraft and its parts Vehicle and its parts Iron and steel products

Proportion (%) 25.79 20.33 12.50 10.17 9.01

According to the proportion of China’s exports to Laos in this year’s exports (shown in Table 3), it can be seen that the most advantageous and most exported are electronics, machinery, spacecraft and their parts, vehicles and their parts, steel products, etc. Electronic products and spacecraft, as a representative of high value-added manufactured freights and capital- and technology-intensive products. The freights that Laos exports to China with greater advantages (shown in Table 2) are mainly low value-added primary products such as processed wood and mineral products. China’s exports to Laos are high value-added products, and technology-intensive products represented by high technology.

3 Freights Organization After the Completion of the China-Laos Railway For the type of trade between China and Laos and the infrastructure of Laos, the freights organization can be divided into the way according to the business model and the type of freights. 3.1

Freights Organization Model According to the Business Model

The freights organization of the business model can be roughly divided into three types as the railway itself marketing center organizes the freights, the third-party logistics enterprise organizes the freights, and both of them organize the freights together. According to the actual situation in Laos, it is recommended to establish a railway freights marketing center based on the China-Laos railway and cooperate with thirdparty logistics companies. • Marketing center organization freights The main freights traded between China and Laos, that is, commodities such as staple cargo, because of the ﬁxed place of production, the cargo owners are relatively ﬁxed. The transportation model is single, and the marketing is not difﬁcult. It is recommended to use the marketing center to organize the freights. The marketing center organizes the freights directly from the railway freights marketing personnel to the unit or individual that produces the freights, including mineral products, wood products, copper and its products, rubber and its products, vegetable oil, seeds, fruit grains and their products and fossil fuels. Production companies, sales companies and other scattered customers.

Research on Freights Organization Strategy

63

The establishment of the freights marketing center of China-Lao railway needs to be based on the source of freights survey in the early stage of railway construction, and the area where the source of freights is more concentrated as a freights marketing center, so as to organize the freights more quickly and efﬁciently. This approach allows the railway to more directly and accurately understand the needs and freights information of each customer. With the close trade between China and Laos, the railway freights marketing center of China-Laos railway needs to be equipped with human and material resources to meet the demand, and the service concept and methods must be continuously improved, closer to the needs of shippers, railway freights. The way that marketing center organizes the freights is an important method for the future freights organization of China-Laos railway. • Third-party logistics companies organize freights For the scattered freights, due to their complicated types, random distribution, and the small volume of single scattered freights, marketing is difﬁcult, so it is recommended that third-party logistics enterprises have to organize the freights. The freights organization of the third-party logistics enterprise can not only improve the efﬁciency, but also reduce the input of manpower and material resources of the railway department. Therefore, cooperation with third-party logistics will also be an important method to solve the problem of the distribution of scattered freights in China and Laos. In the choice of third-party logistics enterprises, it is necessary to comprehensively consider the four aspects of business results, operation quality, service level, service price and select third-party logistics companies that can cooperate with. Since the railway-road transport is the main cargo product of the China-Laos railway, the selected third-party logistics companies need to have the considerable strength to complete the transportation tasks, which plays an important part in achieving “door-to-door” logistics. On the issue of the division of labor between China-Laos railway and third-party logistics companies, China-Laos railway and third-party logistics can establish a vertically integrated logistics alliance. As an upstream enterprise, the China-Laos railway is mainly responsible for the transportation of freights and the efﬁcient transportation of freights on the way. As a third-party logistics enterprise of downstream enterprises, it plays a central role in the organization of the source of freights, and develops a freights cooperative relationship with an upstream enterprise. From the excavation of freights to the whole process of product transportation, the railways will implement integrated cooperation and form a strategic alliance of logistics. 3.2

Freights Organization Mode According to the Type of Freights

• Freights organization of scattered freights For freights with multiple batches and small batches, reasonable and efﬁcient freights organization method is not only the key to shorten the delivery time of railway freights and improve the efﬁciency of railway freights transportation, but also an effective way to meet the needs of shippers and enhance the competitiveness of railway freights transportation [7].

64

Y. Wang et al.

For the China-Laos railway, scattered freights have development potential, to which are needed to pay attention. However, due to its characteristics that distributed distribution, large quantity and a small amount, so it is recommended to transmit to third-party logistics companies to organize the freights. • Freights organization of staple cargo For staple cargo, it is obvious to determine wood products and mineral products as the main staple cargo. Through ﬁeld research, conﬁrming the main distribution points of staple cargo and the source of major staple cargo transportation. The market can rely on the freights marketing center, in the form of public rail transport, and provide containers for transportation. What’s more, it’s necessary to ensure that the staple cargo is sufﬁcient. The guarantee of staple cargo transportation is to ensure the transportation of staple cargo agreements, which is an important means to stabilize the volume of staple cargo. For the freights organization of staple cargo, it is also necessary to strengthen communication with local governments and enterprises in Laos. Through various meetings such as coordination meetings, ﬁeld investigations and visits.

4 Strengthen the Strategy of the Freights Organization 1. Customer classiﬁcation The types of cargo owners are different, and they have a great influence on the choice of shipping methods, railway freights organization and customer service. At present, according to the customer’s trafﬁc volume, railways classify customers into large customers and ordinary customers. The large customer annual transfer more than 1 million tons, and the other is for ordinary customers [5]. For the China-Laos railway, this standard can be changed according to the actual situation. The service flow for large customers can be as shown in Fig. 1. Railway related units can investigate the freights volume of previous years and screen out customers with large cargo transportation demand. These customers will become the major customers of the future railways. Large customers must ﬁrst have long-term and stable transportation demand, which can ensure the transportation volume of transportation enterprises. Secondly, they must have the ability to pay for transportation expenses to provide a guarantee for the economic beneﬁts of railway express transportation enterprises. 2. Establish a regional logistics center Cargo transportation is not only the transportation from the station to the station, but also the process of collecting the source of freights from the source of production to the station and from the terminal to the customer. This is called the distribution of freights. In order to improve the efﬁciency of the freights organization and better cooperation with third-party logistics companies, it is very important to carry out “door-to-door” services and establish corresponding regional logistics centers. Regional logistics centers can fully integrate and utilize the advantages of railways and other

Research on Freights Organization Strategy

65

modes of transportation to build integrated logistics solutions and form a collection and distribution system. The regional logistics center is more attractive to the source of freights, and more efﬁcient distribution of freights.

Fig. 1. The service flow for large customers

5 Conclusion For the China-Laos railway freights transportation, the efﬁcient and high-quality of freights organization method will be the key to reflect the quality of railway freights services. This paper analyzes the main trade commodities between the China and Laos, and based on the principle of the owner’s demand, proposes the freights organization method and strategy of the China-Laos railway. This paper also provides a valuable reference for the freights market of the China-Laos railway, the improvement of railway competitiveness, the creation of a well-established railway freights brand, and the occupation of the Laos freights market. Acknowledgment. This work was supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Eva, S.: Innovation in European Freights Transportation: Basics, Methodology and Case Studies for European Markets. Springer, Berlin (2008) 2. Zhimei, W., Xingchen Z., Bin X.: Railway freights train schedule problem and loading plan for high-value scattered freights. J. Beijing Jiaotong Univ. 40(6), 43–49+56 (2016)

66

Y. Wang et al.

3. Chongyu, L.: Discussion on railway container train operation and freights organization in China. Railw. Freights Transp. 29(1), 27–31 (2011) 4. Cheng, C., Zhiming Y.: Research on freights development and freights organization optimization. Railw. Freights Transp. 31(2), 19–22+50 (2013) 5. Yuzhao, Z., Changfeng, Z., Jianqiang, W., et al.: Strategy and method of expedited cargo source organization in railway transportation under freights organization reform environment. Logist. Technol. 34(19), 1–4 (2015) 6. Niwen, T.: Research on commodity trade between Laos and China. Capital University of Economics and Business (2017) 7. Yuzhao, Z., Jianqiang, W.: Optimization model and algorithm for organization of railway express freight sources. Comput. Eng. Appl. 51(4), 7–10 (2015)

Classiﬁcation of High-Speed Railway Network Transfer Nodes Based on the Improved Gray Whitenization Weight Function Clustering Method Jieying Jiang, Lin Wang(&), and Si Ma School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected], [email protected]

Abstract. To classify the high-speed railway transfer nodes, an improved gray whitenization weight function clustering method based on the entropy weight theory was proposed, the evaluation capability defects and the subjectiﬁcation of weight determination of conventional methods were solved. Main Chinese highspeed railway passenger terminals were divided into preferred network transfer node, sub-selection regional transfer node, general local transfer node, and nonchoice transfer node, which account for about 12%, 20%, 44% and 24% respectively. This classiﬁcation method could lay foundation for the study of high-speed railway transfer mode. Keywords: High-speed railway network Transfer node Gray whitenization weight function clustering Entropy weight theory

1 Introduction With the rapid development of Chinese high-speed passenger railway lines, the shortage of carrying capacity in some mainline railways is becoming increasingly serious. For example, the ratio of Chinese high-speed carrying capacity between Beijing and Shijiazhuaung has reached 93%. And the ratio over 80% accounts for about one third. Under the circumstance of no additional railway infrastructure inputs, transfer mode has attracted particular concern for its simpliﬁed transportation organization, convenient management and great carrying capacity. For a settlement of the shortage of carrying capacity, it is necessary to conduct a study of high-speed railway transfer mode. And the classiﬁcation of high-speed railway transfer nodes is the element task. Transfer nodes selection is the basis of train operation plan. However, the classiﬁcation of transfer nodes is paid less attention. Generally, there are two main methods for classifying the high-speed railway transfer facilities. Firstly, AHP method is used to study the quantitative index of nodes transport distribution capacity and transfer nodes [1, 2]. Secondly, Gray Relational Degree Method is also used to assess the importance of transfer nodes [3]. What’s more, there are other methods for nodes classiﬁcation, for example, K-means clustering method is applied to distinguish degree level of node © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 67–75, 2019. https://doi.org/10.1007/978-3-030-04582-1_8

68

J. Jiang et al.

importance for urban agglomeration [4]. But such methods above all have overlooked subjective effects. An improved method would reduce influence of subjective effects so that a more accurate classiﬁcation and a more speciﬁc study on train operation plan based on transfer mode will be conducted. Gray whitenization weight function clustering method which is applied to synthetic decision, describes the intensity that clustering object belongs to various grayscale [5]. Because of simple calculation and high accuracy, gray whitenization weight function clustering method has widely used in social science and engineering technology. For instance, gray whitenization weight function clustering method is proposed to evaluate the safety level of timber ancient buildings [6] and poor information analysis model of water-saving irrigation projects is proposed based on the center point whitening triangle weight function to analyze the clustering results and optional scheme of the agricultural irrigation project [7]. However, conventional gray whitenization weight function clustering method still uses AHP or Delphi method to determine clustering weight, which is lack of objectivity and evaluation capability. Therefore, in this paper, weight determination method is improved. And a new clustering method based on the entropy weight as well as the gray whitenization weight function is proposed to realize the classiﬁcation of high-speed railway transfer nodes, which would comprehensively evaluate the highspeed railway transfer nodes from passenger demand, network attributes, terminal capacity and social attributes aspect, and solve the problem of evaluation subjectivity, then lay a foundation for study on high-speed railway transfer mode.

2 Gray Whitenization Weight Function Clustering Method Gray whitenization weight function clustering method organizes whitening values of different indexes according to grayscale and evaluation grade in order to judge which grayscale the clustering object belongs to. 2.1

Clustering Grayscale and Clustering Indexes

Considering the influential factors of clustering object, the evaluation system is formed. It is assumed that n clustering objects whose numbers are indicated by i (i ¼ 1; 2; . . .; n), m clustering indexes whose numbers are indicated by j (j ¼ 1; 2; . . .; m) and s clustering grayscales whose number are indicated by k (k ¼ 1; 2; . . .; s) are provided. Clustering index j of clustering object i is indicated by xij . Then according to xij , sample matrix A is obtained. 2.2

Gray Whitenization Weight Function

Gray whitenization weight function is a kind of segmented function whose threshold belongs to 0 to 100. Generally, it is classiﬁed into upper limit measure, middle limit measure and lower limit measure. Whitenization weight function with clustering index j and clustering grayscale k is indicated by fjk ðÞ, which is deﬁned as follows:

Classiﬁcation of High-Speed Railway Network Transfer Nodes

f1 ðxÞ ¼

8 : 70x 7040

8 > 4015

:

1

x 62 ½40; 100 x 2 ½40; 70Þ

ð11Þ

x 2 ½70; 100Þ x 62 ½15; 70 x 2 ½15; 40Þ

ð12Þ

x 2 ½40; 70Þ x 62 ½0 ; 40 x 2 ½15; 40Þ x 2 ½0; 15Þ

ð13Þ

Classiﬁcation of High-Speed Railway Network Transfer Nodes

73

Table 2. Clustering vector of 25 main Chinese high-speed railway passenger terminals Station

Beijing Shanghai Guangzhou Zhengzhou Nanjing Wuhan Hangzhou Changsha Shijiazhuang Jinan Xian Xuzhou Hefei Chengdu Nanchang Guiyang Kunming Fuzhou Shenzhen Haerbin Changchun Shenyang Tianjin Chongqing Nanning

Network transfer node (N) 0.540 0.320 0.443 0.000 0.286 0.429 0.195 0.073 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.171 0.000 0.000 0.000 0.000 0.000 0.000

Regional transfer node (R) 0.213 0.305 0.376 0.353 0.362 0.468 0.325 0.435 0.006 0.137 0.016 0.088 0.057 0.025 0.044 0.011 0.001 0.055 0.145 0.011 0.016 0.016 0.104 0.001 0.016

Local transfer node (L) 0.053 0.099 0.09 0.294 0.224 0.034 0.271 0.048 0.753 0.572 0.855 0.462 0.559 0.601 0.476 0.67 0.507 0.517 0.617 0.136 0.263 0.384 0.313 0.165 0.217

Non-choice transfer node (Non) 0.000 0.000 0.000 0.000 0.000 0.000 0.209 0.361 0.235 0.154 0.113 0.362 0.327 0.348 0.437 0.309 0.492 0.372 0.000 0.842 0.705 0.585 0.479 0.833 0.751

Fig. 2. Classiﬁcation and its ratio of transfer nodes

Clustering result N N N R R R R R L L L L L L L L L L L Non Non Non Non Non Non

74

3.3

J. Jiang et al.

Determination of Clustering Weight and Clustering Coefﬁcient

Combining with Eqs. (5), (6) and (7), clustering weight of four clustering indexes is calculated. w1 ¼ 0:27 (passenger demand), w2 ¼ 0:48 (network attributes), w3 ¼ 0:08 (terminal capacity) and w4 ¼ 0:17 (social attributes). According to Eqs. (8) and (9), gray clustering coefﬁcient is calculated. Clustering matrix is shown in Table 2. Main Chinese high-speed railway passenger terminals are classiﬁed into 4 type. The ratio of 4 type of high-speed railway transfer nodes is shown in Fig. 2. Preferred network transfer nodes including Beijing, Shanghai and Guangzhou, account for about 12%. Sub-selection regional transfer nodes including Zhengzhou, Nanjing, Wuhan et al., account for about 20%. General local transfer nodes including Jinan, Xian, Shenzhen et al., account for about 44%. And non-choice transfer nodes including Changchun, Shenyang, Nanning et al., account for about 24%. From the result above, network transfer nodes locate in Chinese megacity, whose terminal stations have mature facilities and equipment. Regional transfer nodes locate in dense area of Chinese high-speed railway network, whose terminal stations have location superiority as well as larger passenger flow. Local transfer nodes locate in Chinese provincial capital city. Therefore, network transfer nodes are supposed to organize passenger transference and launch departure-arrival trains. By means of regional transfer nodes, a part of overline through multiple unit trains should be converted to intra-line multiple unit trains. Local transfer nodes organize local passenger flow to transit. Besides, non-choice transfer nodes are less taken into consideration when it comes to study on transfer scheme of high-speed railway, but passengers are able to transfer at such terminals.

4 Conclusion There are quantities of influential factors affecting classiﬁcation of high-speed railway network transfer nodes, which leads to fuzziness and hard estimation. Based on gray whitenization weight function clustering method, clustering objects can be classiﬁed according to max gray clustering coefﬁcient. Based on entropy weight, this paper has proposed an improved gray whitenization weight function clustering method and realized the classiﬁcation of main Chinese high-speed railway passenger terminals. In future work, study on high-speed railway network capacity utilization and multiple unit train Scheduling by means of “intra-line and transfer” mode, considering network transfer node and regional transfer node, will be conducted. Besides, through various transfer nodes, multiple unit trains with different speed can be separated in order to improve capacity and efﬁciency of high-speed railway network, which will lay foundation for the study of high-speed railway transfer mode. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702).

Classiﬁcation of High-Speed Railway Network Transfer Nodes

75

References 1. Du, X., Niu, Y., Han, B., Liu, M., Zhu, Q.: Train service planning for passenger dedicated railway line based on analyzing importance of nodes. J. Beijing Jiaotong Univ. 06, 5–10 (2010) 2. Yuan, M., Deng, L., Huang, Y.: Study on railway intermodal transport and transfer node selection. Mod. Bus. Trade Ind. 01, 23–25 (2010) 3. Huang, L., Lv, H., Lin, Y., Wang, W.: Optimization of high-speed train stopping schemes based on node importance. J. Transp. Eng. Inf. 03, 49–57 (2017) 4. Song, X., Wang, X., Li, A., Zhang, L.: Node importance evaluation method for highway network of urban agglomeration. J. Transp. Syst. Eng. Inf. Technol. 02, 84–90 (2011) 5. Li, Y.: Clustering Method in Transportation Field, p. 89. Science Press, Beijing (2014) 6. Guo, X.D., Fu, T.B., Xu, S.: Safety assessment of timber ancient buildings based on gray clustering analytical method. J. Beijing Univ. Technol. 05, 80–85 (2017) 7. Shen, J., Wu, F.: Study on investment decision-making for agricultural water-saving irrigation projects based on gray white function. Acta Agric. Jiangxi 05, 18–21 (2016) 8. Feng, Y., Li, X., Li, X.: Comprehensive evaluation method for the railway safety based on the entropy method and the gray relation analysis. J. Saf. Environ. 02, 73–79 (2014) 9. Liu, J.: Lectures on Whole Network Approach, pp. 129–130. Shanghai People’s Press, Shanghai (2014) 10. Wang, W., Liu, J., Jiang, X., Wang, Y.: Topology properties on Chinese railway network. J. Beijing Jiaotong Univ. 03, 148–152 (2010)

Research on High-Speed Railway Network Effectiveness Based on Theory of Constraints Su Liu, Jieying Jiang(&), and Si Ma School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected], [email protected]

Abstract. Analytical method of carrying capacity based on Theory of Constraints was put forward. The effectiveness of HRN under the existing highspeed railway transportation organization was analyzed. The results show that 37% of the total hub stations are signiﬁcantly capacity limited. The utilization ratio of the high-speed railway network section is 69% on average, and 26% of the section capacity utilization ratio is more than 85%. There is a signiﬁcant imbalance in the utilization ratio of section capacity. The failure of 15%–25% and 33%–47% sections, the efﬁciency of HRN decreases rapidly under intentional attack. After the failure of 50% sections, the change of efﬁciency index of HRN under random attack and intentional attack is relatively stable. On the whole, the intentional attack is more destructive to the structure of high-speed railway network. Keywords: High-speed railway network Effectiveness Theory of Constraints Carrying capacity

1 Introduction Recently, high-speed railway in China has entered an accelerated period of interaction between operation and effectiveness. The capacity utilization imbalance of high-speed railway network (HRN) is becoming increasingly severe. The performance of highspeed railway network carrying capacity distribution is extremely important for the whole network effectiveness. Therefore, how to measure and identify the high-speed rail network effectiveness to optimize transportation organization of HRN becomes an important issue in the HRN. There is a wealth of literatures on the capacity optimization for high-speed railway. Many models were presented to maximize the total capacity. According to different research objects, the high-speed railway capacity is generally divided into passenger station carrying capacity, section carrying capacity and network capacity, covering the three-capacity research scale of HRN. In terms of passenger station and section carrying capacity studies, some studies deﬁne carrying capacity from different perspectives [1–3]. Based on previous research, a new capacity calculation method is proposed [4, 5]. The traditional carrying capacity analysis methods usually include mathematic calculation method, graphic method and simulation method. The accuracy of the three © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 76–84, 2019. https://doi.org/10.1007/978-3-030-04582-1_9

Research on High-Speed Railway Network Effectiveness

77

methods is quite different from the objectives [6]. In the last few decades, signiﬁcant research effort have been devoted to developing analytical models for network capacity. It proposed a conceptual framework of network capacity [7, 8]. And a calculation methods and practical applications is proposed [9–11]. The existing research are mainly related to the carrying capacity of a hub station or a path as the object, ignoring the influence of network characteristics. The few existing studies of network effectiveness and the bottlenecks of HRN are mainly limited to two aspects: carrying capacity and network reliability. In this paper, to further study on the current status of HRN operations, we proposed a method to analyze and calculate the network effectiveness of HRN based on Theory of Constraints, which integrating carrying capacity and effectiveness of HRN, bottleneck identiﬁcation through utilization calculations, robustness and sensitivity of network flow. Bottleneck and sensitivity analyses are conducted to analyze the performance measure with respect to current operational status, ﬁnally, identify the critical roadway components.

2 Network Capacity Problem Base on Theory of Constraints The objective of high-speed railway network capacity base on Theory of Constraints is to determine the maximum attainable flow that a network can carry, which takes into account the capabilities of the passenger station at both ends and the carrying capacity between the sections under the conditions of ﬁxed facilities and devices. The highspeed railway network carrying capacity is the result of coordination and optimization of the point-to-line level of the high-speed railway network system. 2.1

Carrying Capacity of High-Speed Railway Passenger Station

The influencing factors of high-speed trains occupying the throat area of the station are not taken into account during the optimization process. In this paper, carrying capacity of receiving-departure track is the main factor limiting the capacity of high-speed railway passenger station. The carrying capacity of receiving-departure track can be formulated as follows: NRD ¼

MRD ð1440 tT Þ ð1 aLC Þ K tAVE

ð1Þ

Where NRD is carrying capacity of receiving-departure track, MRD is number of receiving-departure track, tT is the total leisure time for stopping train reception and departure within one day and night, aLC is leisure coefﬁcient of receiving-departure track, K is utilization rate. tAVE is an average occupancy time of receiving-departure track which is occupied by one passenger train: tAVE ¼ apas tpas þ atb ttb þ adep tdep þ aavv tavv

ð2Þ

78

S. Liu et al.

Where, apas , atb , adep , aavv are the proportion of passenger trains, which passing through, turn-back, originated and terminal, to the total number of passenger trains reception and departure day and night, respectively. tpas , ttb , tdep , tavv is the occupancy time of receiving-departure track, which passenger trains passing through, turn-back, originated and terminal, respectively. 2.2

Carrying Capacity of High-Speed Railway Section

Removal coefﬁcient method is adopted to analyze the carrying capacity of high-speed railway section. Based on parallel train diagram of high-speed railway, the number of low speed trains is converted into high-speed ones’. The maximum number of highspeed trains is calculated under a certain number of low-speed trains. N is carrying capacity of parallel train diagram: N¼

1440 Tw 60S VI I

ð3Þ

Where, Tw is maintenance time for high-speed railway, I is a minimum time interval between trains spaced by automatic block. S is the length of section. V is the average speed of high-speed trains. The carrying capacity of non-parallel train diagram without low speed trains is formulated as follows: Nw ¼

NA0 eA

ð4Þ

Where, NA0 is the maximum carrying capacity of high-speed trains, eA is the removal coefﬁcient of high-speed trains. The carrying capacity of multiple-speed trains on the same train path is formulated as follows: Nmul ¼ NA þ NB ¼ N NB eB þ NB

ð5Þ

Where, NA is number of high-speed trains, NB is number of low-speed trains, eB is the removal coefﬁcient of high-speed trains. 2.3

Carrying Capacity of High-Speed Railway Section Based Theory of Constraints

If the capacity of the section between the two hub stations is less than the capacity assigned to the section by the hub stations, the carrying capacity of High-speed railway section, according to Theory of Constraints, is the capacity of the section between the two hub stations: Nij ¼ min(Ni ; Nj ; Nijsection ; Nijassigned Þ

ð6Þ

Research on High-Speed Railway Network Effectiveness

79

Where, Ni is the capacity of hub station i. Nj is the capacity of hub station j, Nijsection is the capacity of the section between hub station i and hub station j, Nijassigned is the capacity assigned to the section by hub station i and hub station j. 2.4

The Utilization Ratio of the Carrying Capacity of HRN

The utilization ratio of the carrying capacity is formulated as follows: r¼

Nmul Nij

ð7Þ

3 Example This paper adopted the down direction data of the high-speed railway timetable with 30 hub stations in October 2017. aLC is 0.25–0.35. K is 0.90–0.95. Standard time for all passenger trains at station receiving-departure track is shown in Table 1. When the number of passenger trains passing through the station is less than 10% of the total number of passenger trains reception and departure at the station day and night, aLC is 0.25. For every 10% increase in the ratio, the value increases by 0.01. The carrying capacity of hub station is shown in Table 2. Table 1. Standard time for all passenger trains at station receiving-departure track Train of type Occupancy time (min) 300 km/h 200 km/h Through 3 3 Stop 10 13 Originated 21 22 Terminal 19 19 Turn-back 27 -

Tw is 240. I is 5. eA is 1.1–1.25, eB is 2.5–4. Carrying capacity of High-speed railway section based on removal coefﬁcient method, carrying capacity of High-speed railway section based Theory of Constraints and the actual carrying capacity of Highspeed railway section is shown in Table 3, respectively. It can be seen from Table 3, 11 of hub stations, such as Chengdu, Kunming, Guiyang, Fuzhou and Shangrao etc., are signiﬁcantly capacity limited, which is 37% of the total. The carrying capacity of the section is greatly affected by the capacity limitations of the hub stations at the ends, such as Shangrao-Fuzhou, NanchangFuzhou, Jinhua-Wenzhou, Chengdu-Chongqing, Changsha-Guiyang, GuiyangKunming, Kunming-Nanning, Guiyang-Kunming, Chengdu-Jiangyou, Chengdu-Mt. Emei, etc. Obviously, the phenomenon of capacity limitations are noteworthy, where

80

S. Liu et al.

the sections in the western region connected by Chengdu, Kunming and Guiyang etc. e.g. limited by the carrying capacity of the Guiyang hub, the carrying capacity of the Guiyang-Kunming sections, Guiyang-Guangzhou sections and Changsha-Guiyang sections has been reduced by more than 50%. The utilization ratio of carrying capacity based on the TOC theory is shown in Fig. 1. The utilization ratio of the high-speed railway network section is 69% on average. There is a signiﬁcant imbalance in the utilization ratio of section capacity. 26% of the section capacity utilization ratio is more than 85%, such as Hefei-Nanjing, GuangzhouShenzhen, Beijing-Shijiazhuang, Nanjing-Shanghai, Nanchang-Shangrao and so on. 37% of the section capacity utilization ratio is more than 80%. 63% of the section capacity utilization ratio is more than 70%. 83% of the section capacity utilization ratio is more than 50%. Only 4% of the section capacity utilization ratio is less than 30%. The distribution of High-speed railway carrying capacity presents spatial tropism. On the whole, the capacity utilization ratio of the high-speed railway network in the central and eastern regions of China is higher than that in the western region, due to the differences in passenger demand in the eastern and western regions of China, the imbalance of the density of the railway network and the differences in terrain conditions etc.

Table 2. Carrying capacity of hub station Harbin

Changchun Shenyang Beijing

Tianjin

Shijiazhuang Jinan

Xi’an

Zhengzhou Xuzhou

tT

480

480

480

360

420

480

480

480

480

480

tAVE

20

16

14

21

14

14

14

21

14

12

NRD tT

226.5 Nanjing 420

295.5 Shanghai 420

375 Hefei 480

520.5 Wuhan 480

430.5 300.5 Chongqing Chengdu 480 480

393 231 Hangzhou Jinhua 480 480

370 Wenzhou 480

405 Nanchang 420

tAVE

15

20

17

20

20

22

15

10

16

NRD tT

642.5 430 Shangrao Changsha 480 480

304.5 Guiyang 480

453.5 194.5 Kunming Fuzhou 480 480

235.5 Hengyang 480

540 Guilin 480

283.5 182 Guangzhou Shenzhen 480 480

303.5 Nanning 480

tAVE

10

17

18

22

18

9

13

20

19

19

NRD

283.5

483

252

175.5

288

374.5

291

405

341

375

17

This paper analyzes the effectiveness and changing trend of HRN from the perspective of topological characteristics and operational characteristics based on the study above. The robustness of random attack and intentional attack is calculated with the capacity utilization ratio of HRN. The failure probability of the actual hub node is extremely low, therefore, this paper only takes the change trend of robustness which caused by the failure of high-speed railway section into account. The robustness of High-speed railway network is shown in Fig. 2. As the number of failure sections increases, network efﬁciency continues to decrease. The failure of 0–14% sections, the efﬁciency index of HRN under random attack strategy is lower than that of intentional attack. The failure of 15%–25% and 33%–47% sections, the efﬁciency of HRN

Research on High-Speed Railway Network Effectiveness

81

decreases rapidly under intentional attack. After the failure of 50% sections, the change of efﬁciency index of HRN under random attack and intentional attack is relatively stable. On the whole, the intentional attack is more destructive to the structure of highspeed railway network.

Table 3. Carrying capacity of High-speed railway section QiqiharHarbin

HarbinChangchun

ChangchunHuichun

ChangchunShenyang

BeijingShenyang

ShenyangDandong

Shenyang-Dalian

WuhanNanchang

Nmul

175.5

167.5

163.5

153.5

135

178

178.5

167

Nij

175.5

149.5

146

149.5

135

178

178.5

141

Nactual

52.5 WuhanChangsha

76 NanchangChangsha

78 NanchangShangrao

117 HefeiShangrao

120 ShangraoFuzhou

100.5 ShangraoJinhua

73 Beijing-Tianjin

Nmul

169

190.5

143

177

182

190

208.5

46.5 Beijing-Tianjin (Beijing Shanghai) 210

Nij

163

162.5

143

138.5

75

123.5

208.5

210

Nactual

Nmul

129 BeijingShijiazhuang 164.5

67 ShijiazhuangXi’an 104

130 ShijiazhuangZhengzhou 197

32 TianjinJinan 191

54 JinanTsingdao 157.5

108 NanchangFuzhou 165

135 ShanghaiHangzhou 227.5

135 HangzhouJinhua 187.5

Nij

164.5

84.5

186

191

157.5

62

227.5

160

Nactual

135 JinhuaWenzhou 187.5

135 Wenzhou-Fuzhou

Nmul

135 HangzhouWenzhou 122

117

135 ChangshaHengyang 165

135 Xi’anZhengzhou 180

49.5 JinanXuzhou 187.5

188 ZhengzhouXuzhou 193

133 ZhengzhouWuhan 164

Nij

122

60

117

165

146.5

187.5

184

163

Nactual

109 WuhanChongqing

42 ChongqingChengdu

146 ChangshaGuiyang

135 GuiyangKunming

135 HengyangGuangzhou

65 Hengyang-Guilin

114 GuilinGuangzhou

Nmul

133

158.5

96 ChongqingChengdu (ShanghaiChengdu) 176

177

188

178.5

176.5

159

Nij

112

82.5

92

80.5

86

159

153

123

Nactual

88.5 KunmingNanning

44 NanningGuangzhou

54 Wuhan-Hefei

64 XuzhouNanjing

40 HefeiNanjing

137 NanjingShanghai

97.5 NanjingHangzhou

Nmul

164

137.5

110

187.5

125

198.5

17 NanjingShanghai (BeijingShanghai) 222

Nij

89.5

123

110

187.5

125

198.5

222

179

Nactual

108 GuangzhouShenzhen 176.5

92 Shenzhen-Fuzhou

119 ChengduJiangyou 180

66 ChengduMt. Emei 153

105

151.5

145 GuiyangGuilin 189

206

Nmul

48 NanningGuilin 126.5

Nij

126.5

159

151

85.5

61

61

Nactual

100

151

135.5

75

45

58

179

Taking the capacity utilization ratio and capacity of HRN as the objects, the sensitivity of the carrying capacity change under intentional attack is shown in Fig. 3. It can be concluded that the change caused by the section carrying capacity is more obvious than the change caused by the utilization of the section carrying capacity. The failure of 20% sections, the maximum network carrying capacity caused by the capacity is 10% lower than that caused by the utilization ratio of the high-speed railway network section. For every 10% increase in the failure sections, the value reduces by

82

S. Liu et al.

Fig. 1. The utilization ratio of High-speed railway network carrying capacity

1.2000 1.0000 0.8000 0.6000 0.4000 0.2000 0.0000 0%

20%

40%

60%

80%

Random a ack Inten onal a ack Fig. 2. The robustness of HRN

100%

120%

Research on High-Speed Railway Network Effectiveness

83

8000 7000 6000 5000 4000 3000 2000 1000 0 0%

20%

40%

60%

80%

100%

120%

coeﬃcient of u liza on carrying capacity Fig. 3. The sensitivity of the carrying capacity

7%–9%. The failure of 70% sections, the maximum network carrying capacity caused by the capacity is 70% lower than that caused by the utilization ratio of the high-speed railway network section. This is because the section with high utilization of carrying capacity, in addition to the section with large demand for passenger flow, is more of the section with a small number of actual trains. This kind of section usually runs different speed of passenger trains, severely limiting the maximum carrying capacity of the section, resulting in a large utilization ratio of the section, such as the section from Wuhan to Chongqing. In another case, the carrying capacity of the hub station limits the maximum carrying capacity of the section which is connected to the hub station, and also leads to a large capacity utilization of the section, such as the section from Wenzhou to Fuzhou.

4 Conclusion In this paper, the proposed method analyzes the effectiveness optimization of HRN, and coordination of the carrying capacity of hub stations and sections from a macro perspective. The results show that the high-speed railway network capacity is tight, and. Through the analysis of the characteristics of HRN, it is found that the utilization ratio of high-speed railway network carrying capacity is not balanced and the change of section carrying capacity is sensitive to the high-speed railway network.

84

S. Liu et al.

Acknowledgements. This research was supported by the National Key R&D Program of China (2017YFB1200702). National Natural Science Foundation of China (Project No. 61703351, 71761023), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK0000028-ZF, 2017-RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Peng, Q., Wang, C.: Train Operation Organization, pp. 238–244. China Railway Publishing House, Beijing (2006) 2. Liu, L., Wang, N., Du, W.: A study of network optimization model and algorithm for carrying capacity of station throat. J. China Railw. Soc. 24(6), 1–5 (2002) 3. Arema: Manual for Railway Engineering. American Railway Engineering and Maintenanceof-Way Association, Volume 4- Systems Management (2007) 4. Zeng, J., Liu, J.: Carrying capacity of Beijing-Shanghai high-speed railway by different transport organization patterns. J. Transp. Syst. Eng. Inf. Technol. 12(4), 22–28 (2012) 5. Lv, M., Ni, S., Chen, D.: High-speed railway carrying capacity calculation method. J. Transp. Eng. Inf. 14(1), 19–24 (2016) 6. Chen, T., Ni, S., Huang, Q., Yang, Y.: Model and algorithm for maximum carrying capacity of high speed railway passenger station during peak hours. China Railw. Sci. 36(6), 128–134 (2015) 7. Hu, A., Yang, H.: A series of research on the problem of railway transportation capacity. Railw. Econ. Res. 1, 39–43 (1994) 8. Shi, Q.: Models for rail network system transportation capacity and trafﬁc pathing. J. China Railw. Soc. 4, 1–9 (1996) 9. He, S., Song, R., Dai, X., Yang, Y., Lei, Z.: Study on optimal methods for evaluating the carrying capacity of railroad network. J. China Railw. Soc. 25(2), 5–9 (2003) 10. Lei, Z.: Study on the theory of carrying capacity reliability of railway networks. J. China Railw. Soc. 30(4), 84–88 (2008) 11. Lei, Z., He, S.: Study on the carrying capacity flexibility of railway network. J. China Railw. Soc. 31(1), 26–30 (2009)

Study on the Optimization of Multitransportation Modes in the Surrounding Area of Subway Ya-long Lao1, Lan Liu1,2(&), and Kai-yu Yang1 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. Through the analysis of relevant research at home and abroad, we deﬁne the concept of trafﬁc microcirculation, and point out the problems existing in the current trafﬁc microcirculation. On this basis, we analyze the impact and the characteristics of the subway to the surrounding microcirculation with the goal of optimizing the microcirculation around the subway. And then we develop corresponding optimization goals, strategies and methods. Finally, taking the trafﬁc microcirculation area around the Xipu Station in Chengdu as an example, through the on-the-spot investigation, the problem of trafﬁc microcirculation is clariﬁed, and we put forward the organization optimization suggestion. Keywords: Trafﬁc microcirculation Subway Strategy

Organization optimization

1 Introduction The Athens Charter of 1933 ﬁrst pointed out that the purpose of urban planning is to solve the normal operation of the four functional activities of residence, work, recreation and transportation [1]. In 2017, 26% of China’s urban commuting peaks were in a state of congestion, 55% were in a state of easing, and only 19% were not affected by peak trafﬁc congestion [2]. Faced with the severe problem of urban trafﬁc congestion, most of China’s current urban transportation planning is to optimize the organization of high-grade roads such as highways and main roads, and ignore the optimization of the dense internal minor street of the city to share road trafﬁc [3]. The method of flow, which largely caused the disconnection between urban transportation planning and urban planning, the problem of urban trafﬁc congestion has not been well solved. Although some cities like Beijing have proposed the construction of a trafﬁc microcirculation system, they still neglected the adverse effects of the original function of the damaged microcirculation.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 85–95, 2019. https://doi.org/10.1007/978-3-030-04582-1_10

86

Y. Lao et al.

Since the trafﬁc microcirculation is derived from the microcirculation concept of medicine, the trafﬁc microcirculation can effectively improve the trafﬁc congestion by introducing some trafﬁc jams into the system. Therefore, the trafﬁc microcirculation system has been paid more and more attention by many domestic and foreign researchers. Michael C studied the supply of minor arterials and minor streets, and concluded that travel time is one of the influencing factors of minor street space [4]; Christopher et al. established a design based on the characteristics of trafﬁc routes by designing and managing intersections and circulation routes. A new way of organizing the trafﬁc cycle space replaces the traditional way of trafﬁc circulation [5]. In order to determine the microcirculation sections that need to be optimized for road conditions and trafﬁc organization, a search model for microcirculation bypass routes needs to be established [6]. Faced with the transformation of microcirculation systems in different regions, there are different optimization objectives and constraints, such as minimizing the cost of transformation, minimum saturation of each section of the system, and maximum transformation capacity [7, 8]. Especially in historical blocks, it is necessary to consider protecting historical sites [9]. In addition, the completed trafﬁc microcirculation system also needs to establish a complete evaluation system for the reconstruction effect and the overall system [10, 11]. Because the minor street network trafﬁc has its own internal laws, these laws determine that the microcirculation theory can only play a limited role in trafﬁc planning and minor street network planning [12]. Urban rail transit, represented by the subway, has become an important part of urban trafﬁc as its large capacity, convenience, comfort, speed and punctuality. Therefore, this paper will take the trafﬁc microcirculation system around the subway as the main body, according to the regional characteristics, sort out the characteristics, problems, objectives and optimization strategies of different modes of transportation, and ﬁnally obtain a reasonable organization optimization plan to adapt to the target area.

2 Overview of Trafﬁc Microcirculation Research 2.1

Trafﬁc Microcirculation

(1) Concept of Trafﬁc Microcirculation The microcirculation ﬁrst appeared in the medical ﬁeld and refers to the blood circulation between the arterioles and venules. The internal trafﬁc of the city can be compared to the blood circulation inside the human body, and some minor arterials and minor streets in each sub-area of the city are the micro-arteries and venules in the human body. In view of the similarity between the microcirculation and the urban transportation network, the trafﬁc flow in the city can flow from the urban main road into the microcirculation road system, and then flowing into other main roads, forming a microcirculation of urban trafﬁc, so it can alleviate the trafﬁc pressure on arterials of the city. At present, the generally accepted explanation the trafﬁc microcirculation is that the trafﬁc microcirculation is a road network transportation system which consists of minor streets and some sub-main roads, and roads below the minor street, such as

Study on the Optimization of Multi-transportation Modes

87

streets, roads, and alleys. Since the trafﬁc microcirculation is based on its internal minor street network, it will inevitably interact with the internal ecology, and this paper proposes that the trafﬁc microcirculation system is an organic whole which concludes the road network transportation system consisting of the regional minor street and some minor arterials, roads below the minor street, such as streets, roads, and alleys,as well as the economy, culture, transportation and life in the region. (2) Characteristics of trafﬁc microcirculation In order to properly organize the trafﬁc microcirculation and give a reliable optimization strategy, we must start from the characteristics of the trafﬁc microcirculation itself, understand the difference between it and the conventional trafﬁc, and make corresponding adjustments to its characteristics. To expound the characteristics of the trafﬁc microcirculation, we can start from three aspects: trafﬁc demand characteristics, trafﬁc flow characteristics and functional characteristics. Trafﬁc microcirculation feature is shown in Table 1. Table 1. Trafﬁc microcirculation feature classiﬁcation table Feature classiﬁcation Trafﬁc demand characteristics

Trafﬁc flow characteristics

Functional characteristics

2.2

Features Large raw OD demand Some road service levels are high High connectivity, accessibility, and flexibility Demarcation of road rights and clear trafﬁc priority requirements Slow trafﬁc ratio Small trafﬁc flow and free flow Less disturbance due to fluctuations in mainline trafﬁc flow Diversion trafﬁc Road difference Environmental limitations

Analysis of Existing Problems

As the trafﬁc microcirculation (especially the old city) has not received much attention in urban transportation planning, and the decision-makers lack understanding of the characteristics of the region itself, the following problems still prevail in the current trafﬁc microcirculation system: (1) The structure of the trafﬁc microcirculation system is not perfect In recent years, the trafﬁc construction of domestic cities has vigorously developed the arterial network, while ignoring the construction of low-grade roads has resulted in an unbalanced urban road network structure and the formation of an imperfect microcirculation system, which has intensiﬁed urban trafﬁc congestion to some extent. The imperfect microcirculation system road network leads to poor regional trafﬁc, which not only cannot share the trafﬁc pressure of the arterial network, but also the basic

88

Y. Lao et al.

internal travel demand is difﬁcult to meet. These phenomena are more and more common in the community of large urban centers. (2) Trafﬁc microcirculation optimization strategy pays too much attention to external partial flow function At present, some research and practice pay too much attention to the diversion function of the microcirculation system, ignoring the original functions within the region, it not only encroach on the living space of residents, but also cause exhaust gas, noise pollution, potential safety hazards, and special sites such as schools and hospitals without implemented effective protection, which seriously affects the lives of residents in the trafﬁc microcirculation system. At the same time, due to the internal slow trafﬁc demand and static trafﬁc demand, in the case of excessive attention to external flow, these internal needs will affect the realization of the diversion function, and bring inconvenience to urban residents, but also buried trafﬁc hazards. (3) Trafﬁc microcirculation road lacks perfect trafﬁc management Urban transportation planning and management focus on the trunk road network, and the spatial rectiﬁcation and management of the microcirculation system road network is often neglected. Insufﬁcient information-induced signs and slow trafﬁc facilities have caused a large number of motor vehicles to flow in, pedestrians, bicycles, and motor vehicles have mixed and interfered with each other; The lack of management and restrictions on motor vehicle speeds have caused potential safety hazards, and community restlessness; The lack of static trafﬁc settings has led to the phenomenon of occupation of roads and random parking, which has caused some minor street and streets to lose the basic function of ensuring motor vehicle trafﬁc. In order to relieve the trafﬁc congestion of arterial, the reconstruction or expansion of the microcirculation road network and the lack of reasonable trafﬁc management after introducing the trafﬁc flow into the microcirculation road network are the main causes of the lack of functionality of the microcirculation road network.

3 Analysis of Urban Trafﬁc Microcirculation System Around Subway Station 3.1

The Influence of Subway on the Microcirculation of Surrounding Trafﬁc

Compared with other modes of transportation, subway transportation mainly has the characteristics of public welfare, asset speciﬁcity, high cost, large scale and multiindustry cooperation [13]. According to the speciﬁc analysis of the main characteristics of the subway, it can be known that it will have the following effects on the surrounding trafﬁc microcirculation: (1) Promote residents to use the subway A macroscopic analysis of a subway station can be used as a point of generating and attracting trafﬁc. As one of the important modes of travel for modern residents, the subway has the advantages of safety, comfort, on-time, fast and environmental

Study on the Optimization of Multi-transportation Modes

89

protection. Because of these advantages, coupled with the high accessibility of the subway, the surrounding residents are more inclined to choose to use the subway. (2) Attracting passenger flow in other areas and increasing trafﬁc volume The urban subway network is a huge urban transportation system that can bring the time and space distances of various areas within the city closer. The higher accessibility makes the urban land around the subway station usually respond to the appreciation of the subway [14]. This is called the economic externality of the subway, which can often promote the development of nearby businesses, so that it can gather a large amount of trafﬁc production and increase the amount of the trafﬁc microcirculation. (3) Promote the development of slow trafﬁc within the region The subway station will enable more internal residents to choose subway travel and more external passenger flow, which will increase the amount of slow trafﬁc inside the region and promote the development of slow trafﬁc. 3.2

Characteristics of Urban Trafﬁc Microcirculation Around the Subway

(1) Slow trafﬁc Since the trafﬁc microcirculation area is connected to the subway, the trafﬁc generated and attracted by the subway is slow trafﬁc, and the closer to the subway station, the greater the density of slow trafﬁc. In addition, during peak hours such as commuting, the direction of slow trafﬁc will be more obvious. (2) Motor vehicle flow The main function of the urban trafﬁc microcirculation is to improve the trafﬁc congestion of the city’s main roads, and the appearance of the subway is also to reduce road congestion, so there is a strong trunk trafﬁc diversion capacity in the urban trafﬁc microcirculation area around the subway [13]. However, due to the smooth roads and lack of trafﬁc control in the area, it is prone to the phenomenon that the speed of motor vehicles is too fast, posing a safety hazard. On the other hand, due to the lack of trafﬁc flow, the trafﬁc flow is free, and many uncertain conflict points are prone to occur, resulting in trafﬁc congestion. (3) Public transportation In order to solve the “last mile road” problem of the subway, there are usually many bus lines around the subway station. However, due to the low accessibility within the large residential area, the coverage of bus stations in the trafﬁc microcirculation area is generally low, and the time of walking to the station is long, which is difﬁcult to meet the public transportation needs in the microcirculation area [15]. (4) Static trafﬁc Static trafﬁc can be mainly divided into static trafﬁc of vehicles and static trafﬁc of non-motor vehicles. Since the trafﬁc microcirculation area is connected to the subway, the subway shares part of the motor vehicle travel trafﬁc, so the static trafﬁc of the motor vehicle in the microcirculation area will be more than that of the general trafﬁc microcirculation area; Due to the popularity of shared bicycles, residents’ short-

90

Y. Lao et al.

distance travel is more convenient. A large number of short-distance trips will be attracted around the subway station, so there will be a large number of non-motorized parking needs. 3.3

Optimization Goal and Strategy of Microcirculation Organization Around Subway

The microcirculation of trafﬁc around the subway is to improve the regional road network system, and to provide better service levels for motor vehicles and pedestrians within the region. The service objects can be divided into regional internal trafﬁc flow, inter-regional trafﬁc flow, transit trafﬁc flow according to trafﬁc flow direction. Combined with its characteristics, the optimization goal of the microcirculation organization around the subway is to meet the needs of inter-regional trafﬁc flow and transit trafﬁc flow through effective trafﬁc organization measures on the premise of meeting the needs of intra-regional trafﬁc flow. The optimization strategy of microcirculation organization around the subway can be divided into the following categories: (1) One-way trafﬁc organization optimization One-way trafﬁc organization optimization is a common strategy for trafﬁc microcirculation trafﬁc flow organization optimization. Because of its simple implementation and fast effect, it can also improve vehicle travel delay, improve trafﬁc efﬁciency, reduce the trafﬁc jam on the road and ensure the trafﬁc safety for road sections with poor road conditions. However, one-way trafﬁc will increase vehicle bypass, and commercial activities on both sides of the road will be affected, causing inconvenience to residents who travel, especially those who are not familiar with the road conditions. Therefore, appropriate one-way trafﬁc organization optimization should be carried out according to the actual trafﬁc demand distribution characteristics and the characteristics of the trafﬁc network structure. (2) Intersection steering restriction The intersection steering restriction is also a common strategy for trafﬁc microcirculation trafﬁc flow organization optimization. By restricting the left and right turn of the microcirculation intersection and supplementing the trafﬁc guidance sign, the utilization rate of the microcirculation road is increased to ensure the smooth flow of the trafﬁc. (3) Quiet trafﬁc The tranquility of trafﬁc is a concept derived from Europe and has been valued and used in countries around the world. Its purpose is to enhance the livability of the implementation area, improve pedestrian safety, and mitigate the adverse effects of motor vehicles; its main means is to carry out a series of physical design to reduce the speed of vehicles and restrict the passage of vehicles through vehicles. The purpose of the trafﬁc microcirculation is to provide services for internal residents and to properly serve transit trafﬁc. Therefore, in order to ensure that transit trafﬁc will not have a greater impact on residents’ travel, and also provide a good travel environment for regional residents, consider the design of trafﬁc quietness in the trafﬁc microcirculation implementation area.

Study on the Optimization of Multi-transportation Modes

91

(4) Speciﬁc strategies at the micro level Such as parking planning, division of transportation units and living units in the area, induction of trafﬁc flow, separation of different modes of transportation, etc. These problems are usually related to the original environment and functions in the microcirculation area of the trafﬁc. The influencing factors are complex and difﬁcult to solve through macro-programming. The countermeasures must be given according to the status quo. 3.4

Optimization Method of Trafﬁc Microcirculation Around Subway

According to the above optimization goals and strategies, we can design the optimization process as the following steps: Step 1: Use the oriented graph in the graph theory to represent the road network and divide the segment grade. The node vi represents the feature points on the segment, and the arc lst represents the segment between node vs and vt . Therefore, there is an oriented graph G0 ¼ ðV; LÞ, where V ¼ fvi ji ¼ 1; 2; . . .; ng is the node set, and L ¼ flst ¼ ðvs ; vt Þjvs ; vt 2 V g ¼ flm ; lb g is the arc set. lm ; lb represent the arterial arc set and the minor street arc set. Step 2: With qualitative analysis of each minor street in the trafﬁc microcirculation area, we can divide it into residential living minor streets and trafﬁc minor streets, and reduce the level of transformation of the former. So we can get the new oriented graph G without residential living minor streets. Step 3: Determine the key road sections in arterials based on historical statistics. Step 4: Search for the shortest bypass route in the microcirculation area for a speciﬁc crowded road segment by shortest path algorithm. Step 5: Count all the shortest detours for different congested sections, and form a detour set. Step 6: Simultaneously considering the slow trafﬁc and static trafﬁc around the subway, we will select the key sections of the detour set, and maximize their trafﬁc capacity. At the same time, we should improve slow trafﬁc, public transportation and static trafﬁc, and try to avoid non-interference and illegal parking. Step 7: If the microcirculation bypass route set cannot be generated, check whether there is a broken road. Open the broken road if necessary.

4 Optimization of Trafﬁc Microcirculation Organization Around Chengdu Xipu Station There are many mature residential quarters in the surrounding area of Chengdu Xipu Station. There are many microcirculation systems consisting of minor arterials and minor streets in each community, as shown in the dotted line of Fig. 1. There are ﬁve types of problems: (1) The minor arterial in the microcirculation area lacks trafﬁc control and the trafﬁc flow is chaotic;

92

Y. Lao et al.

Fig. 1. Microcirculation area

Fig. 2. Microcirculation network

(2) Static trafﬁc of motor vehicles is chaotic, which is shown in Fig. 3;

Fig. 3. Chaotic static trafﬁc in microcirculation area

(3) The closure of the community reduces the flexibility and accessibility of the trafﬁc within the area; (4) Lack of trafﬁc guidance devices; (5) The slow trafﬁc volume near the subway station affects motor vehicle travel. In view of the above problems, this paper combines the optimization method to optimize the trafﬁc microcirculation of the surrounding area of Xipu Station, as shown in Fig. 2. The arterials and minor arterials include Hongguang Avenue, Hengshan North Street, Shangjie Xiajie and Xihu Street. Firstly, determine the living area in the area and the vicinity of the subway (inside the yellow dotted line of Fig. 2.), the minor streets in where we can regard as residential living minor streets and reduce their level of transformation. Then, according to the trafﬁc flow characteristics and historical data, locate the congested road section of the surrounding area of Xipu Station, and determine the shortest bypass route, as shown in Table 2.

Study on the Optimization of Multi-transportation Modes

93

Table 2. Frequently congested road and the shortest bypass route Numble Frequently congested road Cause of congestion 1 Hongguang Avenue Large passenger flow 2

Hengshan North Street

3

Xiajie

The shortest bypass route Guoning 1st Street Guoning West Road Large passenger flow Guoning 2nd Street Guoning Middle Street Large passenger trafﬁc Pufa Street Lack of trafﬁc control

In order to make full use of the microcirculation road resources, some internal minor streetes of the system need to be reconstructed, supplemented by scientiﬁc trafﬁc control measures and trafﬁc guidance facilities. The construction of the trafﬁc microcirculation system is as follows: (1) Renovation of Guoning Middle Street and Guoning West Road. Guoning Middle Street is a wide minor street between Diamond Court and Jingxiang Court. It connects Guoning Middle Street and Guoning West Road. The extension of Guoning West Road is consistent with Guoning Middle Street, and form a two-way lane connecting the Hongguang Avenue and Shangjie. (2) Renovation of Guoning 1st Street and Guoning 2nd Street. Guoning 1st Street and Guoning 2nd Street are parallel to each other, and both are connected to Hengshan North Street and Guoning West Road. The Guoning 1st Street is transformed into a one-way street from east to west, and the 2nd street in Guoning is transformed from west to east. These two roads can complement each other. (3) Renovation of Pufa Street. Renovation of Pufa Street will improve the trafﬁc environment of Pufa Street and form a good two-way road. (4) Open the community road. Since the east-west direction of Pufa Street is a broken road, in order to alleviate the trafﬁc congestion in the down street, the broken road should be opened and connected to Xihu Street. (5) Set up signal lights and overspeed detection devices in each minor street to form scientiﬁc trafﬁc control; set up trafﬁc guidance devices in front of the intersection of arterials and minor streets to facilitate the driver’s decision in advance; rational planning next to each minor street and subway station parking space. After the transformation and optimization of the trafﬁc microcirculation system in the surrounding area of Xipu Station, the trafﬁc flow can reach arterials and minor arterials such as Hongguang Avenue and Hengshan North Street through Hongguang, Guoning Middle Street and Guoning West Road, as shown in Fig. 4. This effectively alleviates the trafﬁc congestion on arterial, and at the same time ensures that the slow trafﬁc near the subway station and the living minor street of each community are not affected.

94

Y. Lao et al.

Fig. 4. Optimized microcirculation network

5 Conclusion This paper takes the microcirculation area around the subway as the research object, optimizes its organization, analyzes the deﬁnition, characteristics and problems of the trafﬁc microcirculation itself, and clariﬁes the purpose, strategy and method of organization optimization, with considering the influence of microcirculation on the area. Finally, taking the trafﬁc microcirculation around the Xipu Station as an example, the problem of the microcirculation area was clariﬁed through ﬁeld investigation, and the optimization proposal of the trafﬁc microcirculation organization in the area was proposed. It provides a reference for the organization optimization of other similar types of trafﬁc microcirculation. Acknowledgements. This research was supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Ma, L.: Thoughts on improving the planning and construction of walking and non-motor vehicle transportation facilities. Trafﬁc Transp. 25(02), 1–2 (2009) 2. “2017 China Major Cities Trafﬁc Analysis Report” released. Urban Planning Newsletter (03), 14 (2018) 3. Qiu, F., Chen, F.: The balance of life and efﬁciency—The perfection of urban minor street road network and spatial remodeling. Huazhong Arch. (12), 127–129 (2006) 4. Micheal, C.P.: The supply of Residential Access Streets and Secondary Arterial roads. Transp. Res. Part B 14(1), 121–132 (1980) 5. Christoper, W., David, J., Gautam, A.: Spatial Aspects of Trafﬁc circulation. Transp. Res. Part B 29(1), 1–32 (1995) 6. Zhai, X., Li, Z., Cheng, W., Zhao, M.: Optimization design method of urban trafﬁc microcirculation system. J. Kunming Univ. Sci. Technol. (Sci. Technol.) 35(04), 61–66 (2010)

Study on the Optimization of Multi-transportation Modes

95

7. Shi, F., Huang, E., Chen, Q., Wang, Y.: Single trafﬁc organization optimization in urban microcirculation trafﬁc network. J. Transp. Syst. Eng. Inf. Technol. 9(04), 30–35 (2009) 8. Shi, F., Wang, Y., Chen, Q., Huang, E.: Reconstruction and expansion of microcirculation trafﬁc network and optimization of single-line organization integration. J. China Univ. Min. Technol. 40(02), 327–332 (2011) 9. Wang, Q., Ding, M.: Two-layer programming model for trafﬁc microcirculation optimization in historical blocks. J. Trafﬁc Transp. Eng. 16(03), 125–132 (2016) 10. Huang, E.: Optimization Theory and Method of Urban Road Trafﬁc Microcirculation System Reconstruction and Expansion. Central South University (2009) 11. Zhang, D.: Optimization of urban trafﬁc microcirculation road network based on multiobjective programming. Shenyang University (2014) 12. Zhou, J.: Microcirculation theory and minor street trafﬁc. Urban Transp. 8(03), 41–49 (2010) 13. Zhang, G., Li, J., Guo, J.: The development status and prospect of China’s subway. Shanxi Arch. 36(33), 13–15 (2010) 14. Li, Z., Zhou, S., Wu, S., Dai, Y., Chen, L., Lu, L.: The impact of Nanjing metro on the accessibility of urban public transportation network and land price increase response. Acta Geogr. Sin. 69(02), 255–267 (2014) 15. Wu, X.: Research on the layout method of micro-circulation bus lines connecting urban rail transit. Beijing Jiaotong University (2017)

Coordination Evaluation Model of Metropolitan Rail Transit and Urban Transportation System Yi-Fan Yu1, Jian-Nan Mao1, Lan Liu1,2(&), and Xue-Jiao Xie1 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. There are some problems in the development of metropolitan areas in China, such as the insufﬁciencies of the coordination level and the lack of foresight in planning on metropolitan transportation systems. In this paper, we focus on the research of coordination of metropolitan rail transit and urban transportation system. First, we construct an evaluation index system from both qualitative and quantitative aspects. Then we put forward the weight coefﬁcients based on deviation decision method and then establish a corresponding coordination evaluation model. Meanwhile, an improvement plan will be obtained based on outputted evaluation values. At last, we select the three typical connection points in Chengdu plain city group, evaluate them by the built model and put forward targeted improved proposals through analysis of the results. Keywords: Metropolitan area Rail transit Index system Deviation decision method

Urban transportation network

1 Introduction The national “Thirteenth Five-Year Plan” outline mentions that accelerating the development of urban agglomerations is an important content and an important focal point for optimizing the layout and form of urbanization. However, the development of metropolitan area in China faces some problems, such as the backwardness of the coordination level and the lack of foresight in planning on the metropolitan transportation system. In recent years, many scholars at home and abroad have conducted extensive research on urban agglomeration transportation system. Regarding foreign research on this aspect, Castillo discusses the observability of trafﬁc networks and proposes an improved topology based on existing algebraic methods for solving observability problems [1]. García-Palomares analyzed the relationship between urban expansion and work schedules in the metropolitan area of Madrid, and concluded that the components of the new metropolitan form are related to increasing the number of commuters, the distance, and the model of car use [2]. Panasyuk proposed a solution © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 96–105, 2019. https://doi.org/10.1007/978-3-030-04582-1_11

Coordination Evaluation Model of Metropolitan Rail Transit

97

method and design principle for network optimization of intercity passenger trafﬁc routes based on regional service quality standards, established a passenger trafﬁc model, and proposed road network quality assessment methods and indicators [3]. Kim proposed a trajectory clustering method that ﬁnds spatial-temporal distribution patterns in trafﬁc networks [4]. Regarding domestic research on this aspect, Wu Bing summed up the important characteristics of urban agglomeration transportation system under the background of high urbanization [5]. Jia analyzed the coordinated development of urban agglomeration trafﬁc system, and completed the measurement of trafﬁc development coordination [6]. Guoming has constructed the domestic and foreign typical urban agglomeration transportation network model, and has carried on the contrast analysis about its transportation network characteristic [7]. Taking Changsha-ZhuzhouXiangtan urban agglomeration as an example, Mudan studied the integration of transportation infrastructure construction and urban agglomeration transportation [8]. Wei studied the comprehensive transportation planning of Southern Sichuan Urban Agglomeration [9]. Yuanhui studied the evolution and coordination mechanism of intercity rail transit and urban agglomeration spatial structure [10]. Although many scholars have made great achievements in urban agglomeration transportation system, there is still a lack of theoretical research on rail transit and urban transportation system that they can’t effectively reflect the coordination between them. This article will focus on the coordination problem between rail transit and urban transportation system, establish a corresponding coordination evaluation model, and take Chengdu plain city group as an example.

2 Coordination Evaluation Index System To study the coordination between metropolitan rail transit and urban transport network, it is necessary to consider the connection point of rail transit access to urban transport network and the trafﬁc situation around it. The similar studies can be found in the literature [11, 12], but there is a lack of consideration for trafﬁc conditions around the hub. This paper fully considers this problem. Starting from objective indicators and subjective indicators, the coordination evaluation index system is shown in Fig. 1. (1) the distance between the connection point and the center of the city ðX1 Þ One the one hand, if the distance between the connection point and the city center is too large, it will inevitably cause inconvenience. On the other hand, the distance from the city center should not be too small. Too small distance will cause trafﬁc problems and the layout problems of the urban structure. Due to the rapid development of the current city, for the general large cities, the inner ring area is deﬁned as the downtown area of the city. According to the literature [12], 4–6 km from the city center is the best distance. The distance from the connection point to the city center is classiﬁed as shown in Table 1.

98

Y.-F. Yu et al.

Coordina on evalua on index system Objec ve index

Subway link number

Distance from the center of the city Surrounding highway density

Subjec ve index

Degree of influence on surrounding traﬃc

Intensity of arrival and departure

Conges on degree of surrounding traﬃc network

Convenience through the hub

Average passing me

Safety of the hub

Fig. 1. Coordinated evaluation index system Table 1. Connection point distance grade division table Level A B C D E Distance/km 8

(2) surrounding highway density (X2 ) The greater the density of the highway around the connection point, it is beneﬁcial for the passengers, which means that the coordination is better. The surrounding area is deﬁned as the circle with the radius r at the origin of the connection point. The formula for calculating the density of surrounding highways is shown in (1). P ai l i X2 ¼ i 2 ð1Þ pr where li = the length of the highway i in the region. (3) number of subways and light rail connections (X3 ). Because of the large volume of subways and light rails, it has a great influence on the coordination of rail transit and urban transportation system. The calculation formula for the number of subways and light rails is (2) X3 ¼ n1 þ b n2

ð2Þ

where n1 = number of subways connected to the hub; n2 = number of light rails connected to the hub; b = the scale factor of the light rail (Because the trafﬁc volume of the light rail is less than the subway, it needs to be multiplied by a certain proportional coefﬁcient, generally 0.5.)

Coordination Evaluation Model of Metropolitan Rail Transit

99

(4) hub application strength ðX4 Þ The greater the use intensity of railway hubs is, the less remaining capacity of the hub will be. Here, this paper refers to the ratio of the number N of trains sent by the hub to the number m of railway platforms as the hub to the intensity of application. The calculation formula is (3). X4 ¼

N m

ð3Þ

(5) impact on surrounding trafﬁc (X5 ) After the trafﬁc flow is dissipated from the transportation corridor and the hub of urban transportation, it can enter the urban transportation network through various modes of transportation. When the comprehensive transportation hub is used as the connection point, after the trafﬁc flow enters the junction point from the transportation corridor (mostly the railway), it is generally dissected by means of private cars, taxis, buses, subways, etc. The impact on the trafﬁc around the junction point is mainly to consider the impact on the road trafﬁc, that is, the road trafﬁc flow from the junction point cannot have too much impact on the trafﬁc flow on the original road surface. Its calculation formula is shown in formula (4) X5 ¼

ax ag c z M s þ nx ng C

ð4Þ

where ax = the proportion of people who take a car from the connection point to the total number of people dissolving; ag = the proportion of people who take public transportation from the connection point to the total number of people dissolving; nx = average number of passengers per car; ng = average number of passengers per bus; cz = the conversion factor, the proportion about the number of buses converted into the number of cars, generally take 3 or 4; Ms = unit time trafﬁc flow from junction point to urban trafﬁc network; C = the number of car trafﬁc that can be carried by the urban roads around the junction point. The quantitative analysis of the impact on the surrounding urban transportation network from the convergence of the connection point can reflect the degree of coordination between the transportation channel and the urban transportation network. It is generally believed that if the indicator exceeds 0.4, it will have a greater impact on the surrounding trafﬁc; when the indicator is between 0.2 and 0.4, it has a certain impact on the surrounding trafﬁc; when the value of the indicator does not exceed 0.2, the impact on the surrounding trafﬁc is small [12]. (6) average transit time ðX6 Þ The average transit time refers to the average time consumed by entering the convergence point to leaving the convergence point. It consists of two parts: waiting time and running time. For the case of the transportation hub as a connecting point, the

100

Y.-F. Yu et al.

running time can be speciﬁcally represented by the time spent by the passengers walking in the integrated transportation hub. The waiting time can be embodied as the queue waiting time and waiting time for bus. The formula is shown in Eq. (5). X 6 ¼ ty þ td ¼

l þ td vy

ð5Þ

where ty = travel time through the junction; td = waiting time at the junction; 1 = travel distance at the junction; vy = average travel speed within the junction. (7) the convenience through the hub (X7 ) With the economic development, people’s requirements for convenience are also increasing. For a hub with the large population, people have higher requirements for convenience. The layout and structure of hub should take into account humanity and convenience, the guiding signs in the hub should also be reliable and clear, and more equipment such as escalators should be provided for passengers. (8) hub security ðX8 Þ The security of the hub is a non-negligible indicator. When there is a security risk in the hub, the coordination will be greatly disrupted. The security of the connection points is reflected in the ﬁre protection facilities, the reservation of emergency passages, the delineation of isolation lines, the setting of protection fences. (9) the congestion of the surrounding transportation network ðX9 Þ The congestion level of the transportation network around the hub is closely related to the coordination between the transportation channel and the urban transportation network. When the connecting point is located in a busy trafﬁc area, it not only has great influence on the coordination, but also greatly reduces the capacity of the original transportation network. When there is ample trafﬁc capacity, the coordination is good, causing the trafﬁc volume of the surrounding networks, and even driving the economic development of the surrounding areas.

3 Coordination Evaluation Method 3.1

Standardization of Indicator Data

The set of convergence points is P ¼ jP1 ; P2 ; P3 ; P4 . . .Pm j, and the index set corresponding to the joint point is X ¼ jX1 X2 . . .X9 j. The value of the index Xj corresponding to the connection point Pi is denoted by xij. The matrix corresponding to the convergence point set and the index set is A ¼ xij ði ¼ 1; 2; 3. . .m; j ¼ 1; 2; 3. . .9Þ and called it a decision matrix. This paper uses the efﬁcacy coefﬁcient method [13] to standardize the indicator data. The original efﬁciency coefﬁcient method needs to determine a satisfactory value as its upper bound and an impermissible value as its lower bound. Here the efﬁcacy coefﬁcient method determines a satisfactory value and an impermissible value of the

Coordination Evaluation Model of Metropolitan Rail Transit

101

indicator. For the positive indicator, xg [ xb , for the inverse index xg \xb . By the efﬁciency coefﬁcient method, the original decision matrix A ¼ xij is converted to B ¼ bij ði ¼ 1; 2; 3. . .m; j ¼ 1; 2; 3. . .9Þ. The calculation formula of the efﬁciency coefﬁcient method is shown in Eq. (6). bij ¼

xij xb xm xb

ð6Þ

where xb = unacceptable value of indicator data; xm = satisfactory value of indicator data. 3.2

Determination of the Weight of Evaluation Indicators

The determination of index weights in multi-index evaluation is the core of research on the multi-index evaluation problem. The accuracy of the determination of index weights is closely related to the objective correctness of the coordination evaluation. As analyzed above, for the determination of index weights, the source of weight for subjective empowerment method is based on the experience and analysis of experts and scholars, which inevitably results in the subjectivity and difference of index weights. The data source of the objective weighting method is derived from objective data and has strong objectivity. However, the development time of the objective weighting method is insufﬁcient. It has many deﬁciencies, and the calculation is relatively complicated. In view of the above analysis, this paper refers to the method of determining weights proposed in [11, 14], which is the deviation decision method, and uses this method to calculate and determine the weight of the index system. The principle of this method is as follows. The range of the value of an indicator in all decision-making schemes is not large, indicating that the indicator has less influence on the decision-making of the scheme, and the weight coefﬁcient corresponding to it is smaller. When the value of the change is large, it indicates that the indicator has a greater influence on the decision of the plan. At this time, the corresponding weight coefﬁcient is larger. In other words, the degree of dispersion of the index determines the size of the index weight coefﬁcient, and the two maintain a positive correlation. Therefore, the index weight coefﬁcient can be reflected by dispersion. For the matrix B ¼ bij ði ¼ 1; 2; 3. . .m; j ¼ 1; 2; 3. . .9Þ, I set Lj as the maximum deviation, and the calculation formulas are shown in Eq. (6), (7), (8). Lj ¼ bmax bmin j j

ð6Þ

bmax ¼ maxi bij ; i ¼ 1; 2; 3. . .n j

ð7Þ

bmin ¼ mini bij ; i ¼ 1; 2; 3. . .n j

ð8Þ

102

Y.-F. Yu et al.

Thus, the weight coefﬁcient calculation formula of the index Xj is shown in Eq. (9). Lj wj ¼ P10 j¼1

3.3

Lj

ð9Þ

Calculation of Comprehensive Evaluation Value

In the process of standardization of the number of indexes, the index of the cost index ðX1 ; X4 ; X5 ; X6 ; X9 Þ and the beneﬁt type ðX2 ; X3 ; X7 ; X8 Þ are respectively calculated, making the higher value of bij , the more beneﬁcial to the evaluation result. The weight coefﬁcient of the weight vector W ¼ ðw1 ; w2 . . .w9 ÞT needs more than 0, and meets P9 j¼1 wj ¼ 1. The calculation of the comprehensive evaluation value is (10). ZI ¼

X9 j¼1

wj bij ;

i ¼ 1; 2; 3. . .n

ð10Þ

4 Example Analysis This article takes Chengdu railway station, Chengdu east railway station, and Chengdu south railway station as an example. On one hand, these three connecting points can connect multiple railway lines; on the other hand, there are many deep links with urban trafﬁc networks. Therefore, these three representative connecting points are selected to study the coordination of metropolitan rail transit and urban transportation system. According to Chengdu railway station, Chengdu east railway Station and Chengdu south railway Station, the distance between the hub distance (X1 ), the surrounding highway density (X2 ), the link number of the subway light rail (X3 ), the hub to the use intensity (X4 ), the degree of influence on the surrounding trafﬁc (X5 ), the average passing time (X6 ), the convenience of the passenger passing through the hub (X7 ), the safety of the hub (X8 ), and the congestion degree of the surrounding trafﬁc network (X9 ), these 9 indexes are used to collect and calculate. On one hand, X1 ; X2 ; X3 ; X4 ; X5 ; X6 are the quantitative index, and then the original data are collected and taken into the calculation. On the other hand, X7 ; X8 ; X9 are the qualitative index. Their value is divided into 5 grades: very poor (less than 0.2), poor (less than 0.4), general (less than 0.6), good (less than 0.8), and perfect (less than 1), which are determined by the method of expert evaluation. The speciﬁc results are shown in Table 2 (South, East and North respectively refer to Chengdu south railway station, Chengdu east railway station and Chengdu railway station). According to the data in Table 2, the efﬁciency coefﬁcient method is used to standardize the data. Then the weight of each index is calculated by the deviation decision method. Finally, the evaluation values of the three cohesion points are calculated, as shown in Table 3.

Coordination Evaluation Model of Metropolitan Rail Transit

103

Table 2. Results of index values of each site Connection point X1 X2 X3 South 4.4 0.84 2 East 7.9 0.92 2 North 5.6 0.92 2

X4 35.8 29.4 10.1

X5 0.65 0.83 0.59

X6 0.62 0.88 0.56

X7 0.89 0.91 0.70

X8 0.74 0.76 0.50

X9 0.62 0.57 0.43

Table 3. Results of the coordination evaluation of each site Index Weight South East North

X1 0.096 0.795 0.605 0.695

X2 0.078 0.6 0.8 0.8

X3 0.078 0.35 0.35 0.35

X4 0.099 0.445 0.5625 0.6175

X5 0.141 0.5 0.86 0.48

X6 0.212 0.74 0.96 0.52

X7 0.075 0.90 0.92 0.71

X8 0.078 0.37 0.39 0.13

X9 0.142 0. 7125 0. 5325 0. 475

Comprehensive evaluation value 0.539 0.736 0.648

From Table 3, it is known that X5 (to the surrounding trafﬁc impact degree), X6 (average passing time), and X9 (the congestion degree of the surrounding trafﬁc network) are in the top three of all indicators, which are the main factors affecting the coordination of the site. A very interesting phenomenon is that the index X1 (the distance between the hub distance from the center of the center) is small, only about 10%, which shows that the distance between the rail transit hub and the center of the city has little influence on the coordination.

Fig. 2. Comparison of coordination index values of different connection points

104

Y.-F. Yu et al.

In addition, the coordination index values of the three connection points obtained according to Table 3 are shown in Fig. 2. As shown in Fig. 2, for Chengdu east railway station, in addition to X1 (the distance to the city center) and X9 (the congestion level of the surrounding trafﬁc network), the other indicators are in the forefront of all stations. The results show that the coordination of Chengdu east railway station is better, followed by Chengdu south railway station, and the coordination of Chengdu railway station is ordinary. Therefore, in order to improve the coordination of Chengdu railway station, we should ﬁrst make a targeted transformation. The weight of indicator X5 is relatively large and Chengdu railway station has a large gap with Chengdu east railway station on this indicator, so it is better to improve X5 (the degree of impact on surrounding trafﬁc).

5 Conclusion This paper mainly studies the coordination of metropolitan rail transit and urban transportation system. On the issue of establishing the indicator system, the indicators of trafﬁc conditions around the connection points are innovatively considered. The deviation decision method is used to determine the weight of the indicators more objectively. The representative points in Chengdu plain city group are applied to the model analysis and thus targeted improved proposals are achieved through the analysis of the model’s results. The article basically completed the coordination evaluation of metropolitan rail transit and urban transport network, but there are still deﬁciencies. The indicator system can be further studied, and the travel situation of each traveler can be considered from a microscopic point of view. Acknowledgement. This work is supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Castillo, E., Conejo, A.J., Menéndez, J.M., et al.: The observability problem in trafﬁc network models. IEEE Trans. Intell. Transp. Syst. 9(2), 275–287 (2008) 2. García-Palomares, J.C.: Urban sprawl and travel to work: the case of the metropolitan area of Madrid. J. Transp. Geogr. 18(2), 197–213 (2010) 3. Panasyuk, M.V., Pudovik, E.M., Sabirova, M.E.: Optimization of regional passenger bus trafﬁc network. Procedia Econ. Financ. 5, 589–596 (2013) 4. Kim, J., Mahmassani, H.S.: Spatial and temporal characterization of travel patterns in a trafﬁc network using vehicle trajectories. Transp. Res. Procedia 9, 164–184 (2015) 5. Bing, W., Yanli, W., Zhi, D., et al.: Trafﬁc characteristics of urban agglomeration under the background of highly Urbanization. Urban Trafﬁc 9(02), 67–73+52 (2011) 6. Jia, C.: Study on coordinated development and efﬁciency measurement of transportation system and space in urban agglomerations of China. Southwest Jiao Tong University (2011)

Coordination Evaluation Model of Metropolitan Rail Transit

105

7. Guoming, W., Xiamiao, L., Zhengdong, H.: Comparative analysis of trafﬁc network characteristics of typical urban agglomeration at home and abroad. Comput. Appl. Res. 29 (01), 21–24+31 (2012) 8. Mudan, Y.: Study on the integration of transportation infrastructure and urban agglomeration. East China Normal University (2013) 9. Wei, W.: Comprehensive transportation planning study of Southern Sichuan Urban Agglomeration. Southwest Jiao Tong University (2014) 10. Yuanhui, D.: Study on the evolution and coordination of inter city rail transit and urban agglomeration spatial structure. Jiaotong University, Beijing (2015) 11. Xiaozhu, C.: Transportation channel node coordination analysis. J. Xihua Univ. (Nat. Sci. Ed.) (05), 18–20+28 (2011) 12. Hua, Z.: Research on trafﬁc coordination method and model of railway passenger transportation hub. Southwest Jiaotong University (2010) 13. Pengyu, W., Xiulan, W.: Evaluation of urban land use beneﬁt based on efﬁcacy coefﬁcient method——taking Wuhan City as an example. J. Northwest A&F Univ. (Soc. Sci. Ed.) 8(1), 79–83 (2008) 14. Mingtao, W.: Deviation and mean square error decision making method for determining weights in multi-index comprehensive evaluation. Chin. Soft Sci. (08), 100–101+107 (1999)

Identiﬁcation of Key Nodes and Edges by Importance Ranking and Robustness of Regional Rail Transit Network Si Ma(&) and Ruyi Shen School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected]

Abstract. In this paper, the regional rail transit system is studied by the complex network approach to recognize its important elements, which have great impact on the structure and function of the network. Taking advantage of topological features as indicators, a comprehensive importance ranking algorithm for identifying the key nodes and edges is proposed. Based on the ranking result of Chengdu-Chongqing regional rail transit network, we performed the simulation of network robustness with three attack strategies. Furthermore, we observed that the network can be greatly destroyed while eliminating 6% highest-ranking nodes or edges of the network. These insights can be applied to further protecting the key structure of the network. Keywords: Regional rail transit Robustness

Complex network Importance ranking

1 Introduction As the regional rail transit has gradually formed into a network, the consistent growth of passenger flow increases pressure on this network. In order to ensure the stability and well operation of regional rail transit, it is necessary to strengthen the control of the important stations and lines, and improve the robustness of the network in case of the serious effect from emergency. There are two main methods for measuring the importance of nodes and edges. First, the importance rank of nodes and edges can be directly obtained via network topological features, such as degree centrality [1], betweenness centrality [2], closeness centrality [3], etc. However, any of the indexes can not evaluate the influence of nodes or edges comprehensively and objectively. In [4], a comprehensive evaluation model is established considering a variety of indicators, but the AHP method is more subjective in determining the weights. It is worth noting that [5] ﬁrst adopted the TOPSIS method to make up for the limitation of the subjectivity. Second, the other indirect sorting method is to calculate the degree of network damage after removing nodes or edges, which can also reflect their importance, such as node-deletion method [6] and nodeshrinking method [7]. Node-deletion method evaluates the node importance by measuring the changes of the network connectivity after the removal of the node. Although © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 106–114, 2019. https://doi.org/10.1007/978-3-030-04582-1_12

Identiﬁcation of Key Nodes and Edges

107

this method has the advantage of fast computation, it is limited to connected network, only where the network distance could describe the relative importance of the nodes. Then, the node-shrinking method is proposed that the node importance depends on the increase scale of the network agglomeration by shrinking the edges connected to the node. A disadvantage of this approach is the high time complexity, so it is not suitable for large-scale network. Aiming to quantify the influence of nodes and edges on the network, multiple ranking measures are combined to establish an importance ranking algorithm for nodes and edges to acquire more objective and accurate ranking results. The robustness of the network is deﬁned as its ability to maintain network connectivity after random emergency or targeted attack. Simulating random and targeted attack strategies that have different impact on complex network is a typical way for robustness analysis [8]. Despite the close relationship between effectiveness of targeted attack and the sequence of attack, few studies have been done on the attack sequence, but simply by the order of node degree [9, 10]. To further explore the impact of important elements on network structure and function, we will take the importance ranking results as the input data for robustness analysis, and gain the law of the key nodes and edges of the network.

2 Methodology 2.1

Topology of Regional Rail Transit Network

The selection of the regional rail transit is a collection of rail transit within a regional scope, which correspond to the stations of the Commuter rail, General speed railway, Intercity Railway, and High-Speed Railway systems, and lines which correspond to the railway track between stations. There are three main methods of complex network topology in trafﬁc ﬁeld: LSpace, P-Space and dual method, of which the latter two are typical topology methods of transfer network, and study the transfer property of trafﬁc network. Taking the train flow as the weight assigned to each edge, the basis of this paper is to study the influence of train flow distribution of stations and sections. Considering that L-Space method can reflect the real network in terms of structure and scale, we adopt this method for network construction, and put forward the following topology rules. • Regard each station of the regional rail transit as a node. Take the same transfer station connecting different lines as a single node, but several stations dispersed in the hub are considered as different nodes. • Regard the tracks between adjacent stations on the same line as edges. Several lines between adjacent stations are merged into a single line to study the importance of the nodes, and the same station is divided into different nodes according to the lines to study the importance of the edges. • Take the number of trains between adjacent stations as weights for edges, regardless of the impact of train grade, type, speed and marshalling on weights.

108

2.2

S. Ma and R. Shen

Comprehensive Importance Ranking of Nodes and Edges

At present, the comprehensive ranking method of multiple attribute decision making mainly includes the ideal point method (TOPSIS), the simple linear weighting method and the principal component analysis method etc. Among them, the TOPSIS method is known for objective and reasonable evaluation results, and is widely used as an effective multi-index evaluation method. Therefore, we apply this method to importance ranking of nodes and edges of regional rail transit network. Taking into account both local importance and global importance of the nodes and edges, several topology features are selected as the comprehensive importance decision indexes of nodes and edges of regional rail transit network, as shown in Table 1. The deﬁnitions and calculations of these indexes have been generalize by Háznagy [11]. In addition, the weight of each decision index is calculated by entropy weight method with the data of Chengdu-Chongqing regional rail transit network. Table 1. The decision indexes of importance ranking of nodes and edges Objects Types Node Importance Local Index

Decision Indexes Weights Degree 0.159 Vertex Strength 0.232 Global Index Betweenness of Nodes 0.320 Closeness 0.074 Efﬁciency 0.214 Edge Importance Local Index Degree of End Nodes 0.270 Vertex Strength of End Nodes 0.219 Global Index Betweenness of Edges 0.282 Efﬁciency 0.229

2.3

Robustness Analysis Based on Simulation

1. Robustness Measures The static topological features of complex network theory are mainly suitable for fully connected networks, and some features have displayed inadaptability to the network after being attacked. For example, when a node or edge in a network is attacked to produce an isolated node, the reachability between nodes will be affected. In other words, the distance of node-pair approaches inﬁnity, resulting in the average shortest distance of the whole network tends to inﬁnite. Obviously, the kind of features are not applicable to the robustness analysis. Therefore, this paper studies the robustness of the regional rail transit network by using the relative size of maximal connected sub-graph and the relative size of the network efﬁciency measures. The former index evaluates the changes of the network scale from the structure aspect, and the latter index mainly measures the variation of the network quality from the perspective of function.

Identiﬁcation of Key Nodes and Edges

109

• The relative size of maximal connected sub-graph: The ratio of the number of nodes in the maximal connected sub-graph N 0 to that in the original network N. DM ¼

N0 N

ð1Þ

• The relative size of network efﬁciency: The reflection of the changes in the overall connectivity of the network before and after nodes or edges removal. Making up for the inadaptability of the average shortest distance measure, the relative size of network efﬁciency is equal to the ratio of the efﬁciency before and after the attack on the network. DE ¼

E0 E

ð2Þ

2. Attack Strategies There are two prerequisites for the robustness simulation experiment in this paper. One is that the attack objects is the nodes (stations) or the edges (sections) of regional rail transit network. The node attack means that the targeted node becomes invalid, so does the edges which are directly related to the targeted node. The edge attack just indicates the failure of the targeted edge. Two, there is no protection of the nodes and edges. An attack can lead to the targeted nodes or edges being disabled. Based on the prerequisites, three strategies for robustness simulation of nodes and edges of regional rail transit network are designed. • Random Attack: On the basis of the previous attack, randomly removes the nodes or edges from the current network. Repeat it until the network collapses. • Initial Targeted Attack: Depending on the comprehensive importance ranking result of nodes and edges in the initial network, remove the nodes or edges of from the network in descending order of the importance, until the network collapses. • Dynamic Targeted Attack: Every time the most important node or edge in the current network is removed, the importance of the remaining nodes or edges in the network is recalculated on the basis of the last attack. Repeat it until the network collapses. In addition, the collapse of network indicates that there are no more nodes or edges to be deleted in the current network. That is, all nodes are isolated nodes under the node attack, and all edges are invalid under the edge attack.

110

S. Ma and R. Shen

3 Chengdu-Chongqing Regional Rail Transit Network 3.1

Network Construction

As of March 1, 2018, the Chengdu-Chongqing regional rail transit system consists of 139 stations, 23 lines and 1229 trains per day. The topology of this network is shown in Fig. 1, where the width of each edge represents the volume of the train flow in each section, equaling the weight of the edge.

Fig. 1. The topology of Chengdu-Chongqing regional rail transit network

3.2

Result of Importance Ranking

The TOPSIS algorithm is utilized to acquire the importance ranking list of all railway stations (RS) and sections in the network, as shown in Table 2. By analyzing the results of the network the comprehensive importance ranking, the distribution laws of the nodes and sides with higher importance can be summed up. First, the transfer property of the station is the main influence factor of the node importance. Second, the train flow has a lifting effect on both node importance and edge importance. Third, the importance of end nodes has great contributions to the importance of edges. 3.3

Simulation Result of Robustness

1. Node Attack The attack of nodes in the network is simulated, and the trend of robustness measures is shown in Fig. 2. Viewed as a whole, the descent speed d of robustness indexes under three kinds of attack strategies is d ðdynamic targeted attack Þ [ d ðinitial targeted attack Þ [ d ðrandom attack Þ. When the node failure ratio f is about 40% under random attack, the

Identiﬁcation of Key Nodes and Edges

111

Table 2. The comprehensive importance ranking result of nodes and edges in ChengduChongqing regional rail transit network (Top 15) Node Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Importance Score Node (RS) 0.888 Chongqingbei 0.713 Chengdu 0.709 Hechuan 0.708 Suining 0.683 Dayingdong 0.608 Chengdudong 0.517 Luohuangnan 0.492 Tongnan 0.487 Chongqingxi 0.486 Changshoubei 0.466 Guangyuan 0.460 Qijiangdong 0.460 Deyang 0.441 Ganshuidong 0.423 Tongzibei

Edge Importance Rank Score Edge (Sections) 1 0.711 Hechuan – Chongqingbei 2 0.652 Dayingdong – Suining 3 0.578 Luohuangnan – Chongqingxi 4 0.573 Gaoxing – Chongqingbei 5 0.558 Tongnan – Hechuan 6 0.552 Qijiangdong – Luohuangnan 7 0.543 Changshoubei (C) – Chongqingbei 8 0.542 Chengdu – Dayingdong 9 0.542 Tongnan – Suining 10 0.511 Dayingdong – Chengdudong 11 0.505 Changshou – Chongqingbei 12 0.502 Changshoubei (G) – Chongqingbei 13 0.496 Chongqingbei – Bishan 14 0.492 Ganshuidong – Qijiangdong 15 0.485 Chongqingxi – Chongqingbei

Fig. 2. The trend of robustness measures of the network under the node attack

connectivity of the 40% nodes can be maintained. Two targeted attacks are roughly equal when f is less than 6%. The DE of initial targeted attack plunged to above 0.3, and then the DM drops slowly with f 2 ð10%; 43%Þ. It is worth noting that DM presents a stepwise downtrend at f ¼ 18% and f ¼ 44%, indicating that a key node in the maximal connected sub-graph was removed at this time accounting for the rapid decline in indicators. In terms of dynamic targeted attack, the DM descends to less than 20% when f ¼ 6%. After f ¼ 28%, the DE is almost a straight line, and the network

112

S. Ma and R. Shen

almost collapses. Since then the failure of nodes has little influence on the connectivity of the network. Notice that there are no continuous two sections in the network with f ¼ 65%, which indicates that all the betweenness of the nodes equals 0, so the dynamic targeted attack stops. Consequently, in order to ensure the stable and efﬁcient operation of the network, we must focus on protecting the top 6% nodes in the network. 2. Edge Attack The attack of edges in the network is simulated, and the trend of robustness measures is shown in Fig. 3.

Fig. 3. The trend of robustness measures of the network under the edge attack

Under the edge attack, the order of descending speed of the strategies is still dðdynamic targeted attackÞ [ dðinitial targeted attackÞ [ dðrandom attackÞ. First of all, dynamic targeted attack has the strongest destructiveness and keeps the fastest descent speed. When the edge failure ratio f arrived at 6%, the DM reduced by 56%. After 40% edges failed, the network almost collapsed. Before f ¼ 20%, the DM of initial targeted attacks is higher than that of random attacks because targeted attacks prefer to attack high important edges, which are usually linked to a key transfer node connecting multiple lines. Thus, the network can still maintain a high level of connectivity performance after removing an important edge. When f 2 ð20%; 45%Þ, the destructiveness of initial targeted attack is rapidly rising with a lower DM than random attack. This can be explained that the key transfer node is invalid as a result of all its edges being attacked, so the indexes drops fast. Finally, after f is over 45%, the connectivity of the network under each attack strategy has been reduced to a lower level, so that the changes of indexes among different strategies are not signiﬁcant any more. Compared to the simulation results of edge attack, the robustness indexes of node attack decreases faster and has greater damage to the network. For example, when f reaches 20% under dynamic targeted attacked, the DE for node attack is only 0.1, while the DE for edge attack is more than 0.2. The reason lies in the failure mode assumed in this paper. The node attack means that both the targeted node and the adjacent edges

Identiﬁcation of Key Nodes and Edges

113

will also be invalid, while the edge attack will only make the targeted edge failure. Obviously, the former attack mode is more destructive to the network than the latter. Therefore, the node attack on Chengdu-Chongqing regional rail transit network is more sensitive than the edge attack.

4 Conclusion Taking Chengdu-Chongqing regional rail transit network as an example, the comprehensive importance ranking order of all the nodes and edges is obtained. Then, we simulated experiment of network robustness and draw the following conclusions: 1. Comparing the three attack strategies, the network shows high fault tolerance for random attack, followed by the initial targeted attack, and poor robustness for dynamic targeted attack. It proves that the importance ranking of nodes and edges of the initial network has changed in the process of continuous network attack. Thus, the protection of the elements of the network needs to be carried out dynamically. 2. Based on the results of targeted attack, the robustness indexes of the network decreased most rapidly with about 6% nodes or edges disabled, which are suggested as key protection elements of the Chengdu-Chongqing regional rail transit network. 3. The comparison of node and edge attack simulation indicates that node attack is more destructive to network than edge attack, which provides an insight into what protection should be focused on. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702).

References 1. Albert, R., Jeong, H., Barabási, A.L.: Internet: diameter of the world-wide web. Nature 401(6), 130–131 (1999) 2. Freeman, L.C.: A set of measures of centrality based on betweenness. Sociometry 40(1), 35–41 (1977) 3. Sabidussi, G.: The centrality index of a graph. Psychometrika 31(4), 581–603 (1966) 4. Xu, Y., Gao, Z., Xiao, B., Meng, F., Lin, Z.: Key nodes evaluation with multi-criteria in complex networks based on AHP analysis. In: IEEE International Conference on Broadband Network & Multimedia Technology, pp. 105–109. IEEE (2014) 5. Du, Y., Gao, C., Hu, Y., et al.: A new method of identifying influential nodes in complex networks based on TOPSIS. Phys. A 399, 57–69 (2014) 6. Jiang, Y., Hu, A., Pan, T.: New method for ﬁnding the most vital node in communication networks-node-isolation. Chin. High Technol. Lett. 18(7), 673–678 (2008) 7. Tan, Y., Wu, J., Deng, H.: Evaluation method for node importance based on node contraction in complex networks. Systems Engineering-Theory & Practice (2006) 8. Zhang, J.: Structural Characteristic Studies of Urban Rail Transit Network. Beijing Jiaotong University (2014)

114

S. Ma and R. Shen

9. Gao, P., Hu, J., Wei, G.: Robustness analysis of urban transit network based on complex network with varied weight. Computer Simulation (2013) 10. Du, F., Huang, H., Zhang, D., Zhang, F.: Analysis of characteristics of complex network and robustness in Shanghai metro network. Eng. J. Wuhan Univ. 05, 701–707 (2016) 11. Háznagy, A., Fi, I., London, A., Nemeth, T.: Complex network analysis of public transportation networks: a comprehensive study. In: International Conference on MODELS and Technologies for Intelligent Transportation Systems, pp. 371–378. IEEE (2015)

The Analysis Method of Regional Railway Network Capacity Loss Under Emergent Conditions Pu Wang(&), Yingjie Wang, and Pei-fen Pan Institute of Computing Technologies, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China [email protected], {wangyj,panpeifen}@rails.cn

Abstract. This paper analyzes the trafﬁc flow distribution of the whole regional railway network when the passenger and freight trains reselect the routes which are affected by emergent conditions. By assuming conditions and analyzing variables, the optimized model of a regional railway network capability loss under accident conditions is constructed. With the classiﬁcation of trains affected by disasters, the problem formulation and algorithm are proposed, combined with the idea of minimum cost flow. Finally, the feasibility of the model is veriﬁed by an example. Keywords: Regional railway network Minimum cost flow method

Capacity loss Emergency

1 Introduction Our country have a vast territory with frequent natural disasters, especially in the southwest mountainous area, railway natural disasters also show the characteristics of diversity and wide influence due to its complex topography and physiognomy. In addition, railway accidents and public safety incidents also happen occasionally. These emergencies will result in the decline or even interruption of the capacity of the railway network, which will seriously affect the safety and efﬁciency of railway passenger and freight transportation in China. On the other hand, with the increase of railway mileage, especially high speed railway, the railway network topology has undergone profound changes. At this time, it is difﬁcult to ensure the quality of the solution depended solely on the dispatcher’s experience. Therefore, the scientiﬁc assessment of the change of regional network under the condition of disaster is not only conducive to the risk management of railway, but also beneﬁcial to the uniﬁed and coordinated command of train operation by the railway transportation management department. Scholars at home and abroad have done a lot of research and achieved fruitful results in terms of transportation network capacity and theory. On the basis of the study of the urban road trafﬁc network user equilibrium model, Shefﬁ (1985) proposed the optimal model of the railway network. The objective function of the model is to minimize the total travel time. Considering the constraints of various trafﬁc modes, subsystem equipment capacity, consumable resources, effectiveness of rolling stock © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 115–123, 2019. https://doi.org/10.1007/978-3-030-04582-1_13

116

P. Wang et al.

and transportation cost, Morlok et al. (1981, 1999) took the OD requirement as a vector set, and established a mathematical model to study the capability of integrated transportation system. Clarke (1995) constructed the optimal model of the railway network system with a variety of container logistics and multiple railway competition, and estimated the effect of line transportation capacity and delay on the flow. Shi (1996) studies the transport capacity of the road network system by means of reverse thinking and CIS Research. Starting with the deﬁnition and model of transportation demand, a multi-objective decision model of road network trafﬁc path is put forward, and the model of transportation capacity is given under a certain trafﬁc flow path scheme. Lei (2006) proposed a deterministic trains assignment model for the effective use of transport capacity in road network system. Based on the genetic algorithm of K short circuit, the calculation method of transportation capacity of road network system is given, and the reliability and flexibility index of railway network transportation capacity are studied. Zheng et al. (2011) put forward the concept of the reliability of railway network capacity, constructed the calculation model of the related road network capacity and the calculation method of the reliability of the railway network, and designed a practical example to calculate the reliability of the road network under the different OD demand level. At present, for the study of regional road network capacity, most adopt static methods and deterministic models, while static methods and uncertain models are relatively few. Moreover, the research on the change of regional road network capacity under such conditions as natural disasters and railway accidents is relatively scarce. Through the collection of existing literature, it is found that only Chu et al. (2015) has evaluated the loss of road network capacity in mountain areas under disaster conditions, but it does not consider the limit conditions of station capacity, and the topological distance of each section of the line is set to 1. The accuracy of the results is also open to discussion. In this paper, the evaluation of the loss of railway network capacity under emergent conditions is helpful for the transportation decision-makers to grasp the overall transport capacity of the road network, and to provide some reference for the railway transportation management department to formulate the train operation organization scheme, thus improving the railway transport efﬁciency under the emergent conditions.

2 An Overview of Regional Road Network Capacity Loss In the railway transport organization, there are two forms of railway transport capacity, namely, railway capacity and railway transport capacity. The railway passing capacity refers to the maximum amount of trafﬁc (the maximum number of trains or the logarithms of the train) that can be passed by a variety of ﬁxed equipment in a railway section under the conditions of a certain type of rolling stock and organization of train operation, it is related to the ﬁxed equipment, the type of rolling stock and the method of trafﬁc organization. Railway transport capacity refers to the maximum volume of freight transported (the maximum tonnage of goods) within a unit time under certain conditions of equipment and transport organization. The capacity of regional road network is the maximum transportation capacity that the railway network system

The Analysis Method of Regional Railway Network Capacity Loss

117

actually completed in unit time under the conditions of transportation demand, manpower, trafﬁc organization level, train flow path distribution, ﬁxed equipment and mobile equipment. Therefore, the capacity of the regional road network is related to the facilities of the line, the type of rolling stock, the OD passenger/cargo flow, the train diagram and other factors. The occurrence of emergent events (natural disasters, war, etc.) caused the coordination of lines, transportation facilities and facilities to be destroyed, and it is bound to reduce the capacity of railway network. At this time, reasonable adjustment of train flow path is the most effective way to improve railway network capacity. Regional network capacity loss should be evaluated according to the principle of comparison between “normal condition” and “emergent condition”. Under the conditions of emergent events, the capacity of the lines and stations in the regional road network will be lost to a certain extent. However, due to the adjustable of the trafﬁc flow path and network effects, the loss of the capacity of the equipment can’t be simply equated with the loss of the capacity of the road network, for example, when the train affected by an emergent incident detours other routes to complete passenger and freight transportation, it can be considered that emergencies do not affect the transport capacity of the road network. The analysis shows that the degree of loss of regional road network capacity under emergency is essentially to calculate the transportation demand affected by unexpected events, that is, the proportion of the capacity of the railway network must be stopped after the transportation organization and the adjustment of the trafﬁc flow path to “normal” regional railway network capacity. Under the conditions of emergency, the transportation demand loss in the regional road network is mainly manifested by the train volume that must be stopped because of the too large broad cost of the path adjustment (time, distance, cost, etc.) or the limit of the station and the line capacity.

3 Mathematical Model for the Capacity of Regional Railway Network Under Emergent Conditions 3.1

Problem Description

Passenger- freight mixed-operation mode is adopted by the regional railway network. High-grade passenger trains and special goods train are preferred for path selection. The trains in operation stop at the nearest station, and the de-starting train re-selects the path instead under emergent conditions. 3.2

Assumption

1. The passenger and goods train flows between stations, train grades, and the loss of line and station capacity in the network under emergency are known. 2. No adjustment is made to the normal section of the train diagram. Roundabout trafﬁc is adopted in the emergency section with a certain generalized cost (time, distance, cost et al.).

118

3.3

P. Wang et al.

Problem Formulation and Solution

3.3.1 Variable Description A regional railway network can be abstracted into a graph G ¼ ðV; E Þ which consisting of a set of nodes V and a set of edges E. The set of station can be expressed as V ¼ ðv1 ; v2 vn Þ, the set of section can be expressed as E ¼ ðe1 ; e2 em Þ. eij is section between station i and station j. lij is length of the section, i; j 2 V. Be is the carrying capacity constraint of the e th section in the network, e 2 E. Nv is the carrying capacity constraint of the v th station in the network, v 2 V. Zloss is the loss of regional railway network capacity under emergency. Z is the carrying capacity of regional railway network under normal conditions. e is deduction coefﬁcient. ae is loss ratio of section capacity. bv is loss ratio of station capacity. ak;e i;j is Boolean logic variables of

k;e kv the arc-route, if arc e is on the K th short route, ak;e i;j ¼ 1, otherwise, ai;j ¼ 0. dij is Boolean logic variables of the station-route, if station v is on the K th short route, kv kk dkv ij ¼ 1, otherwise, dij ¼ 0. xij is Boolean logic variables of the flow-route, if the train kk flow is on the K th short route, xkk ij ¼ 1, otherwise, xij ¼ 0. k is properties of train flow. k ¼ 1, passenger flow; k ¼ 2, freight flow. mkij is number of train flow between station i and station j under emergency condition. dij is the j th train flow in class i.

3.3.2 Problem Formulation In order to ensure the maximum passenger and freight transportation, it is allowed to make roundabout transportation within a certain range to reduce the impact of network capacity under emergency condition. Therefore, with the objective of minimizing the loss of regional network capability, the optimization model is formulated as follows: min Zloss ¼

Z

X X i;j2V;i6¼j

k

! e

k

mkij

=Z 100%

ð1Þ

3.3.3 The Constraints Constrained by station capacity. Under emergent conditions, the situation that the stop of the operating trains at the nearest station, and the adjustment of affected train will result in the capacity of the station exceeding the limit. Therefore, the volume of passenger and freight trains through the station is constrained by station capacity: X XX i;j2V;i6¼j

k

k

kv ek mkij xkk ij dij bv Nv 8v 2 V

ð2Þ

The Analysis Method of Regional Railway Network Capacity Loss

119

Constrained by section capacity. When adjusting the transport organization plan under emergent conditions, the total number of trains passing through the line e is constrained by section carrying capacity: X XX ke ek mkij xkk ð3Þ ij aij ae Be 8e 2 E i;j2V;i6¼j

k

k

Constrained by Path uniqueness. The adjusted train flow can only select one feasible route in the road network. X k

xkk ij 1 8i; j 2 V; i 6¼ j; 8k 2 f1; 2g

ð4Þ

Constrained by operational status. If the distance or time of the roundabout or detour is too long, generally take outage or stop measures. Cijk indicates the distance or time of the bypass: f lkij Cijk 8i; j 2 V; i 6¼ j; 8k 2 f1; 2g

ð5Þ

Constraint of logical: xkk ij 2 f0; 1g

3.4

ð6Þ

Solution Approach

3.4.1 Algorithm The route selection is carried out by using the algorithm of minimum cost flow, with the priority of the train. After selecting the train flow according to the level, the OD node of the train is used as the original and terminal of the network. The train flow f between the two nodes is taken as the flow value, and the arc is undirected edge. Then route selection. An accompanying flow increase loop Gf is constructed with accompanying f in the transportation network. Check for negative loops in the augmenting network Gf . If there is no negative loop, it is the minimum cost flow. Otherwise, adjust the increment of the augmenting network Gf in the presence of a viable flow. The minimum cost flow path can be obtained after multiple iterations. Calculate whether the generalized cost of the route is within the allowable range, and determine whether to keep the train route. By this cycle, the route adjustment of all affected trains can be solved. 3.4.2 Solution Process Step 1: Determine where the accident occurred, calculate the proportion of section capacity loss ae and he proportion of station capacity loss bv .

120

P. Wang et al.

Step 2: Analyze the trains affected by the accident, including the operating trains and trains in plan. For the operating train, the original station is changed to the nearest the point on the topology, and the terminal station is unchanged. For the train in plan, no changes will be made. Step 3: According to the requirements of the transportation organization, the affected train flow should be graded, the trains of the same level are in the same set, D1n ¼ fd11 ; d12 ; d1n g. Then, the largest train flow dij from the highest level train flow set Dm is selected. Step 4: Selected the route for the graded train flow. When fij is greater than the section capacity, fij ¼ fij 1. Repeat step 4, until the train flow dij is assigned to a route, or the train flow fij ¼ 0 corresponding to the train dij . Step 5: Deducting the capacity occupied by the train flow from the relevant sections and stations in the network. If the capacity of a section is 0, the section is disconnected. If the capacity of a station is 0, the section connected to the station is considered to be disconnected. Step 6: Select the second largest train flow from the same train set, and go to step 4. If all the trains in the same level are searched already, then the sub-critical train set is searched. The algorithm ends until the entire train flow is completely searched or the whole network is not connected.

4 Numerical Simulations The topology structure of a railway network in a region of Southwest China, as shown in Fig. 1, shows that the passing capacity of each line is known as a determined value. As a result of the disaster, the section between the station B and the station F has been completely damaged,resulting in the section passing capacity of 0, but other lines are not affected, that is, the capacity is constant. At the same time, all trains through the section B - F will be affected, and the affected train will be collected at the nearest station. After the disaster, the number of trains affected in the railway network is shown in Table 1. Among them the trains through station A and I are high class passenger trains and the freight trains are of the same grade.

Fig. 1. Topology diagram of a regional railway network

The Analysis Method of Regional Railway Network Capacity Loss

121

Table 1. Influence of disaster on train operation in railway network Section of Train pairs/ passenger trains (pairs/d) Affected A—I 3 A—F 10 A—G 2 Not A—L 9 affected A—I 4 Note: The letters in the ﬁgure represent stations,

Through section of freight trains A—I A—F A—G A—C A—E the same below.

Train pairs/ (pairs/d) 10 8 6 10 7

The generalized cost is represented by the topological distance of every section in the network, and the residual capacity and generalized cost of the line in the area are shown in Table 2. Table 2. Residual capacity and generalized of lines Sections A—D D—E B—F H—I J—C B—C J—K A—L

Residual capacity 10 15 0 15 3 4 19 5

Generalized cost 1 4 3 2 2 3 3 3

Sections A—B A—J F—G E—G K—C C—H G—I

Residual capacity 7 18 14 13 16 18 10

Generalized cost 1 2 1 2 1 2 3

For the limit of station capacity in the regional network, the virtual edge method can be used to transform the limit of station capacity into the limit of section capacity. For the convenience of calculation, the limit condition of station capacity is not considered for the example. The generalized cost constraint is that the cost of adjusting the train running path will not exceed 3 times of the cost before adjustment (Table 3). Table 3. Train routines and transport capacity after adjustment Train operation adjustment plan Plan 1

Type of train passenger trains freight trains

Direct section A—I A—F A—G A—I A—F A—G

Train Route A—B—C—H—I A—D—E—G—F A—D—E—G A—J—K—C—H —I A—B—C—H—I —G—F A—J—K—C—H —I—G

Train flow 3 10 2 10 1 2 (continued)

122

P. Wang et al. Table 3. (continued) Train operation adjustment plan Plan 2

Type of train passenger trains freight trains

Direct section A—I A—F A—G A—I A—F A—G

Plan 3

passenger trains

A—I A—F A—G

freight trains

A—I A—F

Plan 4

passenger trains

freight trains

A—G A—I A—F A—G A—I A—F A—G

Train Route A—B—C—H—I A—D—E—G—F A—D—E—G A—J—K—C—H —I A—B—C—H—I —G—F A—J—C—H—I —G A—D—E—G—I A—D—E—G—F A—J—K—C—H —I—G A—J—K—C—H —I A—J—C—H—I —G—F A—D—E—G A—D—E—G—I A—D—E—G—F A—J—K—C—H —I—G A—J—K—C—H —I A—B—C—H—I —G—F A—D—E—G

Train flow 3 10 2 10 1 2 3 9 2 10 3 1 3 9 2 10 3 1

When unexpected events cause B-F damage in the regional network and the train was discontinued, train redistribution is carried out. Through the above model, 4 kinds of adjustment plans are calculated, which can adjust 28 pairs of trains to the new path to continue the passenger and freight transportation. One is to adjust 15 pairs of passenger trains and 13 pairs of freight trains, the other is to adjust 14 pairs of passenger trains and freight trains respectively, and two adjustment schemes are included in each case. Considering the type of train and the minimum of generalized cost, we choose plan two as the best option. The loss ratio of the train flow reflects the loss degree of regional network capacity under emergency. The deduction coefﬁcient of the passenger train is set to 1.3, and the calculation result is (0 1.3 + 11)/(28 1.3 + 41) = 14.21%, that is, the capacity loss of the railway network affected by the unexpected events is about 14.21%.

The Analysis Method of Regional Railway Network Capacity Loss

123

5 Conclusions In this paper, the algorithm proposed can intuitively analyze the transportation organization method of the roundabout or detour that the corresponding to its generalized cost value (time, distance, cost of the route adjustment, etc.) while selecting the train route. Further, analysis capacity loss from the perspective of complex networks can take into account the overall effect of sections and stations on the network. Acknowledgement. The research was supported by China railway corporation of emergency preplan compiling and application research (2017F001) and Railway beidou satellite navigation application platform overall plan and platform prototype development (2017X006-F).

References Shefﬁ, Y.: Urban Transportation Network: Equilibrium Analysis with Mathematical Programming Methods. Prentic-Hall Inc., Englewood Cliffs (1985) Morlok, E.K.: Determining the Capacity of a Transportation System, 1981 Annual Report of the Institute of Applied Systems Analysis, Ministry of Research and Technology, Federal Republic of Germany, Calogne, pp. 253–260 (1981) Morlok, E.K., Riddle, S.M.: Estimating the Capacity of Freight Transportation Systems: A Model and Its Application in Transport Planning and Logistics, Transportation Research, no. 1653 (NAS-NRC), pp. 1–8 (1999) Clarke, D.B.: An examination of railroad capacity and its implications for rail-highway intermodal transportation. Ph.D. Dissertation of the University of Tennessee, Knoxville (1995) Shi, Q.Zh.: Models for rail networks system transportation capacity and trafﬁc pathing. J. China Railw. Soc. (04), 1–9 (1996) Lei, Zh.L.: Research on the theory and calculating methods of railway network carrying capacity. Beijing Jiaotong University (2006) Zheng, Y.J., Zhang, X.C., Xu, B., Wang, L.L.: Carrying capacity reliability of railway networks. J. Transp. Syst. Eng. Inf. Technol. 11(4), 16–21 (2011) Chu, Y., Ma, S., Liu, S.: Study on evaluation methods of capacity loss for railway network in mountain area under disaster condition. Railw. Transp. Econ. 03, 64–68 (2015) Dempere-Marco, L., Melcher, D.P., Deco, G.: Effective visual working memory capacity: an emergent effect from the neural dynamics in an attractor network. PLoS ONE 7(8), e42719 (2012) Zhao, X.G., Du, Y.P., Zhu, A.H., Zhang, Y., Wang, X.H.: Research on generating for train paths in trafﬁc interruption conditions on china railway network. Procedia Eng. (137), 772–776 (2016) Pavlov, D.: Effect of corrosion layer on phenomena that cause premature capacity loss in lead/acid batteries. J. Power Sources 48(2), 179–193 (1994) Lin, B., Qiao, G.: Iterative algorithm of railway network empty cars distribution based on restriction of route capacity. China Railw. Sci. 29(1), 93–96 (2008)

Technical Measures of Controlling Train Headway on High-Speed Railway Yuhua Yang1,2,3, Shaoquan Ni1,2,3, Minghui Wang1,4, and Guangyuan Zhang1,3(&) 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected], [email protected], [email protected], [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 Chongqing - Guizhou Railway Co., Ltd., Chongqing 400014, China

Abstract. The In accordance with the standard calculating method, the influence factors analysis of high-speed railway train headway is carried, and the technical measures of cutting down the train headway is proposed. Based on CRH380AL EMU with the speed of 300 km/h, the contrast of the train tracing interval of Beijing-Shanghai high-speed rail before and after the control is obvious, which is calculated by the traction calculation simulation software. The result proved the probability of setting the train headway as 4 min on BeijingShanghai high-speed rail. In addition, the experiment examine the technical measures of cutting down the train headway is scientiﬁc and reasonable. Keywords: High-speed railway

Train headway Technical measures

1 Introduction Train headway is one of the influence factors on high-speed railway carrying capacity, is also the major approach to improve the efﬁciency of organization. At present, China’s high-speed train headway operate by the standard of 5 min, is different from the standard of 3 min for 300 km/h high speed railway standardized by Speciﬁcation of High Speed Railway Design. In previous research, Wei simulated the aim-interval control mode curve and calculate the headway [1]; Zhang put forward a computing method of high-speed railway train headway [2]; Tian calculated the train headway theoretically and analyzed the main influence factors [3]. Geng focused on the relationship of Neutral Zone and headway [4]; Zhang simulated train on Beijing-Shanghai high-speed railway and proposed several ways to shorten headway [5]; Yang discussed the calculation method of train headway, and analyzed the measures of shorting headway from optimizing station design, EMU design, signal and blocking equipment [6]. Howlett deeply studied the train control optimization strategy, and developed the corresponding train optimization control assistant system to simulate the train operation © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 124–132, 2019. https://doi.org/10.1007/978-3-030-04582-1_14

Technical Measures of Controlling Train Headway

125

process and get the train interval time [7]. Ling combined with the on-site joint debugging experience to conduct an in-depth study on the standard test method for train headway between high-speed trains [8]. Yang studied the braking capacity calculation of EMU with the goal of the train tracking interval, and the six-stage model of EMU braking was introduced in the paper [9]. This paper will analyze the influence factors of headway on high-speed railway, put forward synthetically technical measures to shorten headway. Besides, a simulation of train on Beijing-Shanghai high-speed railway will be conducted, the validity and effectiveness of synthetical measures will be proved.

2 Influence Factors of High-Speed Railway The operating environment of the EMU on high-speed railway is different from that of the ordinary-speed railway. So as the factors affecting headway. Through the analysis of actual process of high-speed train, it is concluded that the factors mainly include train control and interlocking system, ATP speed monitoring mode curve, train traction and braking performance, station and the line conditions, the length of block section, neutral zone, and driver’s operation. The factors affecting the headway of high-speed railway trains are shown in the Fig. 1 below.

Fig. 1. Factors affecting high-speed train headway

126

2.1

Y. Yang et al.

Train Control System and Interlocking System

Different train operation control systems correspond to different train operation modes. High-speed railway trains use CTCS-2 and CTCS-3 train control systems, all operating in a continuous primary speed curve control mode. In different systems, vehicle-toground information transmission mode is different. In CTCS-3, the GSM-R wireless communication technology is used to realize communication between vehicle and ground. In CTCS-2, the ZPW-2000 track circuit and active or passive transponders are adopted. Transmission method affect the time, and have different effects on headway. The station takes the time required for picking up and dispatching, which is affected by the signal, interlocking equipment. 2.2

ATP Speed Monitoring Mode Curve

The onboard ATP control mode curve is affected by the relevant train control system. The high-speed railway train runs according to the ATP monitoring curve. The actual running curve has a certain safety margin compared with the monitoring mode curve. The calculation method of ATP control mode curve set by different signal manufacturers can be divided into two types: European standard calculation method and Japanese calculation method. The difference of parameter setting causes the difference between train monitoring braking distance and train control curve setting, and the train headway. 2.3

Train Traction and Braking Performance

The time required for train accelerating from zero to target speed reflects the traction performance. When exiting the station, train accelerates out of the station until reaching the maximum allowable speed. The time required decelerating from the allowable speed to zero reflects the braking performance. When entering the station, train decelerates with the brake condition until it stops. The inbound braking and outbound traction processes are converted into the time, which are important parts of headway. Therefore, for different EMU equipment, better traction braking performance of the train, smaller headway related to the braking deceleration and the traction acceleration; conversely, worse traction braking performance is associated with longer headway. 2.4

Station and Line Conditions

Factors such as ramps, curves, line speed limits, station throat lengths, and ballast speed limits have direct impact on train operation. The train is subjected to additional resistance on a long ramp. It is difﬁcult to reach the maximum allowable speed, causing headway longer. The speed is also limited by ATP speed curve. Train needs to decelerate in advance into speed limit section, and then accelerate to normal speed after passing through it, resulting in train headway increasing. The train entering and leaving the station is affected by the lateral limit speed of the ballast and the train control system. Due to the speed, the time of running through station and outbound throat is long, thus the departure and arrival headway are long. At

Technical Measures of Controlling Train Headway

127

the same station, the throat area of the outermost channel of the station is longer than which near the station line. The impact on departure and arrival headway is obvious. In the process of entering and leaving the station, the lateral speed limit of the switch directly relates to the headway. The larger the number of turnouts, the higher the lateral speed limit value, which is more conducive to the train clearing the throat quickly, thus shortening the departure and arrival headway of high-speed railway trains. The number of turnouts in the throat area increases, the average speed of the station through the throat improves, but it will increase the length of the throat area of the station and prolong the time of the train passing through the throat area. In order to minimize the train headway, the reasonable allocation of the number of turnouts in the throat area and the length of the throat area of the station needs further research. 2.5

Length of Block Section

The length of train blockage zone is closely related to the interval tracking time of the trailing train. Moreover, the length of the occlusion zone on the high-speed railway line is different. When the current train runs in a long occlusion zone, the interval of the trailing train is longer. In the composition of the train departure headway, after the departure of the train, the station will handle the corresponding operation for the following train. Therefore, the occlusion section length setting is related to the length of the train departure headway. The proximity of the station is similar to the train arrival headway. 2.6

Neutral Zone

The train cannot obtain the electric energy of the traction train from the neutral zone, and can only operate under inert conditions. If the train enters the neutral zone during the acceleration process, even if there is a neutral zone in the departure, the train departure headway is extended. In addition, the different setting forms of the neutral zone have different lengths, and the impact on the train headway is also different. 2.7

Driver Operation

The operation of trains in lines and stations is not only directly controlled by train control system, but also directly controlled by the driver. The calculation of the train headway is based on the actual running curve of the train. Therefore, the closer between the driver control curve and the monitoring curve, the smaller the headway. But in practice, it is redundant on the basis of ensuring safety, and more reservations on the train running speed, resulting in redundancy in the actual check value of the train headway.

128

Y. Yang et al.

3 Technical Measures of Controlling Train Headway 3.1

Optimize Design of Line and Station

Under the premise of ensuring the safety, the length of key occlusion zone on the train headway should be shorten, and the interval headway will be shortened. Reasonably shorten the length of the departure, so that the station can handle the outbound approach of the train as soon as possible for late train, so as to shorten the departure headway. “Short neutral zone” can be used as much as possible, and should be placed away from the throat area of the station, to reduce the impact on arrival and departure headway. If meets the requirements of operation, optimize station throat. For stations with multiple connection directions and a lot of transmission lines, consider designing several sub-ﬁelds to reduce the throat length, and use large-scale turnouts to increase the lateral passing speed, thus shortening arrival headway and departure headway. 3.2

Optimize Control System and Signal Equipment

Some trains use the guidance mode and CTCS-2 partial monitoring mode. It is recommended to upgrade to CTCS-3 full monitoring mode to improve the train speed limit value, so that the train can fully utilize the entry and throat settings. The signal for stopping the train from the side line and starting from the side line should be set according to the speed limit value of 80 km/h. It is recommended to change the station switch conversion mode from “series” to “parallel”, to shorten the operation time of equipment and the additional time of operation, thus shortening the train departure headway and arrival headway. 3.3

Improve Train Traction and Braking Performance

Improving the braking performance of the train from the mechanical equipment level, especially the EMU of the running speed above 300 km/h, could shorten the train braking distance, and effectively control the arrival headway. Similarly, improving the traction performance of the EMU can be effective shorten the train departure headway. It is recommended to optimize the equipment braking parameters, standardize the ATP braking curve deceleration calculation model, further verify the safety margin and try to make a uniﬁed conﬁguration, to improve the vehicle control capability and shorten the headway of high-speed railway trains. 3.4

Improve Driver’s Operation

Under the premise of safety, it is recommended driver control the vehicle as possible as close to ATP control mode curve, especially passing through throat; after the train leaves, the driver should speed up as soon as possible to shorten departure headway. In summary, the headway of high-speed railway train is affected by many factors, and the formulation of control methods should also be considered in combination with multiple elements. Under the premise of ensuring the safety of train and meeting the

Technical Measures of Controlling Train Headway

129

demand, with the aim of improving the running speed and shortening the various working hours, the train headway can be controlled from basic hardware facilities, software equipment, driving operation and driving organization management. In the actual transportation organization, it is necessary to combine the various control situations of stations and lines for compressing the train headway, or to adopt different methods according to different interval requirements.

4 Train Headway on Beijing-Shanghai High-Speed 4.1

Calculation Method for Train Headway

Train headway is a shortest time between two tracking trains in same direction on automatic block sections [10]. In Train Headway Checking Method for High Speed Railway, train headway I is divided into four categories, train interval headway Iinterval , train arrival headway Iarrival , train departure headway Ideparture , train headway when passing through the station Ipass . The calculation method in theory is shown as follow. Iinterval ¼ 3:6

Lbraking þ Lprotect þ Lblock þ Ltrain þ tadd vinterval

Ideparture ¼ 3:6 Iarrival ¼ 3:6 Ipass ¼ 3:6

Lsign þ Lblock þ Ltrain þ tdeparture vdeparture

ð1Þ ð2Þ

Lbraking þ Lprotect þ Lthroat þ Ltrain þ tarrival varrival

ð3Þ

Lbraking þ Lprotect þ Lblock þ Ltrain þ tpass vpass

ð4Þ

Thus, the train headway on high-speed railway is: I ¼ max Iinterval ; Ideparture ; Iarrival ; Ipass

4.2

ð5Þ

Basic Parameters

1. EMU model is CRH380AL, ATP type is 300T, and train length Ltrain is 403 m. 2. The train safety protection distance Lprotect is 60 m in station and 110 m. 3. The CRH380AL EMU is composed of 16 vehicles, the distance from the outbound signal to train parking gauge Lsign is 65 m. 4. Adopted CTCS-3 control system. 5. The maximum speed limit of line is 300 km/h; lateral limit speed of ballast is 80 km/h. 6. Simulated driver controlling, the traction coefﬁcient is set to 0.9.

130

4.3

Y. Yang et al.

Beijing-Shanghai High-Speed Train Headway Check

Combined with the actual parameters of the Beijing-Shanghai high-speed railway, the train departure station of Beijing South Railway Station is controlled by the speed limit, and the station throat is passing by at 45 km/h. The train arrival headway and departure headway are calculated when passing the longest throat. Interval headway is calculated when the previous train is in the longest occlusion zone. Set the driver to keep the remaining of 15 km/h when train passing turnouts. The results are shown in Table 1.

Table 1. Train headway on the railway from Beijing to Shanghai before and after controlling Station

Arrival headway Before After Shanghai Hongqiao 251 242 Kunshan South 240 219 Suzhou North 246 229 Wuxi East 253 233 Changzhou North 245 225 Danyang North 227 210 Zhenjiang South 254 239 Nanjing South 250 247 Chuzhou 246 220 Dingyuan 242 230 Bengbu South 282 244 Suzhou East 251 230 Xuzhou East 268 243 Zaozhuang 255 234 Tengzhou East 252 241 Qufu East 246 238 Tai’an 252 236 Ji’nan West 284 247 Dezhou East 268 243 Cangzhou West 245 234 Tianjin South 245 231 Langfang 256 238 Beijing South 289 248

Departure headway Before After 189 178 169 159 175 165 171 160 165 155 150 142 176 164 171 165 172 162 168 160 185 172 163 153 208 194 163 153 149 141 174 164 182 170 194 180 174 163 176 165 171 161 202 188 257 233

Passing headway Before After – – 149 149 155 155 153 153 148 148 142 142 154 154 155 155 151 151 152 152 163 163 149 149 169 169 151 151 145 145 151 151 154 154 162 162 158 158 153 153 152 152 163 163 – –

Interval headway Before After 177 177 138 138 146 146 141 141 139 139 143 143 140 140 128 128 134 134 144 144 143 143 139 139 131 131 143 143 153 153 135 135 154 154 127 127 139 139 135 135 139 139 119 119 – –

Improved the speed limit of train entry and exit station to 75 km/h, shortened the length of the station from the occlusion zone by 200 m, reduced the length of the throat zone by 100 m * 300 m, and increased the throat speed limit of Beijing Nanjing Station to 80 km/h, then the train headway after partial control is as shown in Table 2 below.

Technical Measures of Controlling Train Headway

131

Comparing the simulation results, it can be seen that the key to affect the headway of Beijing-Shanghai high-speed trains is Iarrival and Ideparture . Measures such as improving the speed limit of the throat in the station, improving the driver’s ability to control the train, shortening the distance between the occlusion zone and the throat of the station can effectively control the train headway of the Beijing-Shanghai high-speed train. The proposed multi-management control method can shorten the train headway of high-speed railway trains in combination.

5 Conclusion The train headway of high-speed railway is an integrated performance of infrastructure equipment and transport operations. The headway is affected by many factors as mentioned above. Diversiﬁed measures can effectively shorten the headway. The coordination and efﬁcient combination of multiple systems such as lines, stations, communication signals, and train control systems are necessary ways to ensure the minimum headway at stations and sections. To effective controlling of train headway, it is necessary to combine the other aspects such as the design of the throat area of the station or the management of the transportation organization in further study. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351, 71761023), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK0000028-ZF, 2017-RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Wei, F.H., Wu, Q., Liu, L.: Simulating calculation and optimization design of the tracing time interval of trains in aim-interval control mode. J. Transp. Syst. Eng. Inf. Technol. 7(3), 105–110 (2007) 2. Zhang, Y.S., Tian, C.H., Jiang, X.L., Wang, Y.B.: Calculation method for train headway of high speed railway. China Railw. Sci. 34(05), 120–125 (2013) 3. Tian, C.H., Zhang, S.S., Zhang, Y.S., Jiang, X.L.: Study on the train headway on automatic block sections of high speed railway. J. China Railw. Soc. 37(10), 1–6 (2015) 4. Geng, J.C.: Analysis of influence of contact points of high-speed railway contact network on train tracking interval. Chin. Railw. 1(10), 7–10 (2011) 5. Zhang, Z.H.: Discussion on the tracking interval of Beijing-Shanghai high-speed train. Railw. Commun. Signal Eng. Technol. 13(03), 8–11 (2016) 6. Yang, C.H.: Discussion of shortening train headway under one-brake control mode. J. Transp. Eng. Inf. 8(1), 20–24 + 30 (2008) 7. Howlett, P., Milroy, I.P., Pudney, P., et al.: Energy-efﬁcient train control. IFAC Proc. Vol. 26(2), 1081–1088 (1993)

132

Y. Yang et al.

8. Ling, X., Yang, W.T., Amp, T.: Study on method standardization of high-speed railway tracking train interval test. Railw. Stand. Des. 59(10), 23–27 (2015) 9. Xin, Y., Lin, S., Shao, J., et al.: Calculation method for braking capacity of high speed EMU based on train tracking interval time. China Railw. Sci. 34(06), 99–104 (2013) 10. Peng, Q.Y., Wang, C.G.: Railway Trafﬁc Organization, 1st edn. China Railway Publishing House, Beijing (2015)

Signal Timing Optimization of Isolated Intersection for Mixed Trafﬁc Flow in Hanoi City of Vietnam Using VISSIM Xuan-Can Vuong1,2, Rui-Fang Mou1(&), Hoang-Son Nguyen1, and Trong-Thuat Vu2 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 630031, China [email protected], [email protected] 2 University of Transport and Communications, Hanoi, Vietnam

Abstract. Recently, motorcycle explosion has become a big challenge for sustainable development in Hanoi city of Vietnam. Hanoi has been facing increased annoying issues, including congestions, environmental pollutions and accidents, leading to increase in delay and long queue formation at signalized intersections. Hence, the purpose of the study is to demote delay at isolated signalized intersections under mixed trafﬁc flow with motorcycle-dominated. The optimum cycle length model is re-calibrated to the Hanoi trafﬁc condition based on the Webster’s method. Delay is selected as the performance index in those models using VISSIM. By taking a typical intersection in Hanoi as a case study, the optimal scheme is identiﬁed to reduce average delay by 23.9%, which further veriﬁed the modeling capability of the proposed method. Keywords: Signalized intersections VISSIM

Mixed trafﬁc Optimization

1 Introduction Presently, trafﬁc situation in Hanoi city of Vietnam is characterized by a great volume of motorcycles and a road network with limited technical infrastructure [1]. In Vietnam, the terminology ‘‘motorcycle’’ used in this proposal refers to motorized two-wheelers, including mopeds, scooters, and normal motorcycles which their engine capacity generally ranges from 50cc to 150cc [2]. Motorcycle ownership has been increasing at more than 10% per year for the last two decades, so the ownership rate is over 600 motorcycles per 1,000 people [3]. Motorcycle is shared about 80% of travelling, and is also one of the main cause of trafﬁc problems, such as congestions, pollutions and accidents. According to DOT of Hanoi [4], in 2016, there were 41 locations congested on rush hours, and almost of them are related to intersections, especially at-grade signalized intersections. The average time of trafﬁc jams in Hanoi is from 45 min to 60 min per day, and time loss and increasing of fuel costs per year was estimated to be 22 trillion VND (equivalent to 1.0 billion USD) [5]. Furthermore, trafﬁc fatality rate is extremely high, about 8 deaths per 100,000 population and motorcycles were involved © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 133–139, 2019. https://doi.org/10.1007/978-3-030-04582-1_15

134

X.-C. Vuong et al.

in 46.8% of the total accidents in 2016. Trafﬁc is one of the main causes to environmental pollutions in Hanoi. A report of GreenID showed that the PM2.5 index in Hanoi is ﬁve times higher than the World Health Organization’s annual average level [6]. To solve above problems, Hanoi has made efforts to improve trafﬁc quality of road network, in particular trafﬁc at signalized intersections, such as flyover bridge, trafﬁc signs, pavement markings, extension of geometric, and so on. However, in fact, almost intersections are pre-timed signal-control systems with simple control schemes. Moreover, the design speciﬁcations for signal intersections are not detailed enough, so these control schemes are often determined according to the personal experience of trafﬁc engineers, affecting the efﬁciency of trafﬁc flow operation. Under conditions of implementation of adaptive signal control in Vietnam still face many obstacles, such as complicated trafﬁc flow, complex technique, large investment etc., one of effective measures, that does not require a large budget to relieve trafﬁc congestion and improve trafﬁc safety at intersections is to optimize signal timings. A lot of recent studies have been considered to evaluate best way of signal timing optimization at intersections based on theory and/or software as well as practice. However, very little of those methods can be widely utilized in different trafﬁc conditions, especially the trafﬁc dominated by motorcycles like Hanoi. Hence, the objective of the study is to establish an optimized signal timing plan at isolated intersection under mixed trafﬁc condition according to historical data based on a ﬁxed scheme of phase and phase sequence. Delay is selected as the performance index in those models using VISSIM.

2 Research Methodology Optimal-delay cycle length by Webster is now being used in many countries at an isolated intersection. It is obtained using Eq. (1) [7]. Co ¼

1:5L þ 5 1Y

ð1Þ

Here Co is optimum cycle length P (s); L isP total lost time per cycle (s); Y is sum of the critical flow ratio for phases, Y ¼ yi ¼ qi =si ; qi is trafﬁc volume of the critical lane in phase i (PCU/h); si is saturation flow of that critical lane (PCU/h). The Webster’s method is quite consistent with the homogeneous trafﬁc flow, but this method is applied directly to mixed trafﬁc flow dominated by motorcycles like Hanoi. Therefore, it is difﬁcult to fully reflect the characteristics of trafﬁc stream, so it is necessary to adjust. In that direction, Cuong [8] proposed equation of optimal cycle length under motorcycle-dominated flow based on Webster’s method, but there is limited practical veriﬁcation. Thus, this study combines Cuong’s proposal with HCM’s method to optimize signal timing plan at isolated intersection in Hanoi using VISSIM micro-simulation as an evaluation environment, as follows.

Signal Timing Optimization of Isolated Intersection for Mixed Trafﬁc Flow

135

Optimal cycle length by Cuong [8] for mixed trafﬁc flow as in Eq. (2). Co ¼

1:5L þ 5 P 1 qmc i þ 1 f smc i

qcar i scar i

ð2Þ

car are trafﬁc volume of motorcycles and cars, respectively; smc Where qmc i and qi i and are homogeneous saturation flow of motorcycles and cars, respectively; f is an adjustment factor depending on the trafﬁc models at trafﬁc signals, and f ¼ 0:8 0:9 for mixed trafﬁc condition as in Fig. 1. Homogeneous motorcycle saturation flow is the value of 11,000 motorcycle-unit/h/3.5 m, which was researched by Hien and Frank [9], and homogeneous car saturation flow is determined based on HCM [10].

scar i

Fig. 1. The trafﬁc models of trafﬁc flow at trafﬁc signals

After determining the optimum cycle length according to Eq. (2), based on the principle is to obtain minimum overall delay, the total effective green time should be distributed among the different phases in proportion to their Y values to determine the effective green time for each phase [7]. Whereby, it is easy to obtain the effective green time for each phase by the following equation. X mc 1 qmc qcar 1 qi qcar i i i gi ¼ ðC LÞ : þ car = þ car f smc f smc si si i i

ð3Þ

Where C is cycle length used (usually obtained by rounding off C0 to the nearest ﬁve seconds, and to satisfy the minimum green time for pedestrians in the directions according to HCM [10]), gi is effective green time in phase i(s). In order to evaluate the effect of proposed method, the trafﬁc micro-simulation technology was used. Presently there are number of simulation tools used in order to understand trafﬁc flow trend, such as VISSIM, PARAMICS, AIMSUN, SYNCHRO, TRANSYT-7F and so forth. Among them VISSIM shows its superiorities with the ability to simulate multi-modal trafﬁc flow modeling. Both private transport involved motorcycles and public transport can be analyze under complex geometries. In addition, the various parameters required in VISSIM can be calibrated accommodation with trafﬁc condition. The case study was conducted using the simulation software package VISSIM [11]. Most parameters of motorcycle behavior were calibrated using the data of Minh [12]. According to Minh [12], in undivided streets with mixed trafﬁc in Hanoi, the average free flow (or desired) speed of motorcycles is about 21.1 km/h. Some additional parameters required in VISSIM were also calibrated, such as look-ahead

136

X.-C. Vuong et al.

distance and average standstill distance by Duy [13], minimum lateral distance by Van et al. [14], and some remaining such as deceleration rates (1 m/s2 per distance) had to be assumed. For car and bus, the distribution of desired speed of these four wheelers in the models is assumed to be similar to that of motorcycles, but with a shorter range of variation. Due to the lack of available data of car driving behavior in mixed trafﬁc conditions like Hanoi, all other parameters for driving behavior of cars and buses were taken as the default values in VISSIM. This study uses the calculated mean value based on 5 random seeded VISSIM simulation.

3 The Case Study of Hanoi City The case study analyses a typical isolated intersection (Nguyen Chanh-Mac Thai Tong) in Hanoi, capital of Vietnam. The plane graph was screenshot from Google Map, shown in Fig. 2. The intersection has four arms where each of them has two lanes in each direction. The road was deﬁned as downtown motorized road, with 3.5 m of width per lane. A geometric layout of the intersection is given in the Fig. 3.

Fig. 2. Screenshot of intersection

Fig. 3. Geometric layout of intersection

Signal Timing Optimization of Isolated Intersection for Mixed Trafﬁc Flow

137

Trafﬁc volume of the four sections of this intersection was collected during morning peak hour from 7 am to 8 am in working day in April 2018 by videography and manual counting. The PCU factor of types of vehicle is taken from TCVN 4054 [15]. The values were shown in Table 1. Table 1. Trafﬁc volume in four sections (per hour) Approach Direction Westbound Left-turn Straight Right-turn Eastbound Left-turn Straight Right-turn Northbound Left-turn Straight Right-turn Southbound Left-turn Straight Right-turn

Bicycle Motor-cycle Car Mini-bus Bus Truck PCU 3 327 91 5 0 5 212 7 2037 607 14 20 17 1340 0 124 39 0 0 1 79 2 217 71 3 0 5 155 0 1331 430 22 22 19 976 0 120 37 0 0 2 78 2 122 32 0 1 6 87 4 388 108 0 0 8 249 2 104 44 0 0 0 76 0 292 117 10 0 6 240 1 394 133 6 0 14 298 1 138 71 3 4 4 139

The intersection now adopts a two-phase signal control plan, as shown in Fig. 4 Table 2.

Fig. 4. Phases of trafﬁc signal circle at the intersection in morning peak hour

Table 2. Optimization signal phases at the intersection Parameters Optimal cycle length (s) Cycle length (s) Green time (s) Phase 1 (North) Phase 1 (South) Phase 2 (West) Phase 2 (East) Yellow (s) All-red time (s)

Before optimization After optimization – 66 31 31 29 29 3 0

54 60 15 15 35 26 3 2

138

X.-C. Vuong et al.

With the calibrated parameters as mentioned above, 10 simulation runs were carried out, as in Table 3.

Table 3. The evaluation results between before and after optimization Indexes Before optimization After optimization Avg. delay (s) 40.7 31.0 LOS D C

Fifty vehicles were observed to measure the delay time at the intersection during the morning peak using video playback. The mean were taken with a mean of 39.6 s (SD = 5.3 s). As shown in the table above, the average delay of existing signal timing scheme using VISSIM is 40.7 s which has little difference from the observed values of 39.6 s, the errors were within 3%, therefore, the overall performance of the calibrated model was considered to replicate the real trafﬁc conditions. After optimization, the delay time of existing scheme was reduced from 40.7 s to 31.0 s (equivalent to 23.9%), accordingly, level of service (LOS) of intersection was also improved.

4 Conclusions To judge the effective of signal timing optimization, the delay time which is typical feature of trafﬁc flow quality, is presented by using VISSIM to compare with existing signal timing plan. The results of applied signal timing optimization at Nguyen ChanhMac Thai Tong intersection showed that using optimized trafﬁc signal timing can help to reduce average delay and to improve the service level of intersection. Therefore, currently, in Hanoi, in addition to the improving geometry of intersections, the optimization of signal timing plan according to actual trafﬁc conditions is very necessary. Acknowledgements. This research work was supported by the National Key R&D Program of China (No. 2016YFC0802209).

References 1. Sohr, A., Brockfeld, E., Sauerländer, A., Melde, E.: Trafﬁc information system for Hanoi. Procedia Eng. 142, 220–227 (2016) 2. Minh, C.C., Sano, K., Matsumoto, S.: Characteristics of passing and paired riding maneuvers of motorcycle. J. East. Asia Soc. Transp. Stud. 6, 186–197 (2005) 3. Tuan, V.A.: Mode choice behavior and modal shift to public transport in developing countries - the case of Hanoi City. J. East. Asia Soc. Transp. Stud. 11, 473–487 (2015) 4. DOT of Hanoi: General Report. Hanoi (2016) 5. Chung, P.H.: The result of collection and compilation of 10 SUTI index in Hanoi City. ESCAP, Hanoi, Vietnam (2017)

Signal Timing Optimization of Isolated Intersection for Mixed Trafﬁc Flow

139

6. VietNamNet Bridge: How serious is air pollution in Vietnam? 18 April 2017 (2018). http:// english.vietnamnet.vn/fms/environment/176546/how-serious-is-air-pollution-in-vietnam-. html. Accessed 7 Apr 2018 7. Nicholas, J.G., Lester, A.H.: Trafﬁc and Highway Engineering, 4th edn. Cengage Learning (2009) 8. Cuong, D.Q.: Trafﬁc signals in motorcycle dependent cities. Technische Universitt Darmstadt, Germany (2009) 9. Hien, N., Frank, M.: Different model of saturation flow in trafﬁc dominated by motorcycles. In: Proceedings of the Eastern Asia Society for Transportation Studies, vol. 6 (2007) 10. TRB: Highway Capacity Manual (HCM), Washington D.C (2000) 11. PTV AG: VISSIM 4.30 User Manual, Karlsruhe, Germany (2007) 12. Minh, C.C.: Analysis of motorcycle behaviour at midblocks and signalised intersections. Nagaoka University of Technology, Japan (2007) 13. Duy, T.Q.: Study for calibrate parameters of car following model which is applied to simulate the motorcycle trafﬁc by trafﬁc simulation method. HCMC University of Technology (2012) 14. Van, H.T., Fujii, S., Schmocker, J.D.: Upgrading from motorbikes to cars: simulation of current and future trafﬁc conditions in Ho Chi Minh City. J. East. Asia Soc. Transp. Stud. 8, 335(2008) 15. MOT of Vietnam: The Highway Design Standards (TCVN4054-2005), Hanoi (2005)

Railway Timetable Diagnostic Analysis Based on Train Operation Data Changan Xu1,2,3, Shaoquan Ni1,2,3, Shengdong Li1,2,3, and Dingjun Chen1,2,3(&) 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected], [email protected], [email protected], [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China

Abstract. This paper presents a novel railway timetable diagnostic analysis framework based on train operation data. Firstly, we deﬁne the concept of train deviation and give a new deﬁnition of train delay. Secondly, we propose the delay high-incidence section, station and train identiﬁcation methods On the basis of deviation-based visualization technology. Finally, we take ChengduKunming railway line (from Puxiong-Xichangbei section) as an example to verify the proposed method. The results show that: (1) Our method can effectively identify stations and sections that are prone to delays; (2) Cargo trains are more likely to occur delay than passenger trains; (3) There is a strong linear correlation between arrival delay and departure delay. Keywords: Railway timetable Diagnostic analysis Train operation data Statistical methods

Train delay

1 Introduction Railways occupy an important position in China’s transportation mode. By the end of 2017, China railway mileage reached 127,000 km. Of which nearly 20% are highspeed railways, and 80% are conventional railway line. In 2017, the railway company transport 3.084 billion people. And the total freight transport volume exceeds 3 billion tons. Timetable is the basis of ensuring the effective and safe operation of the trains on such a large-scale railway network. It determines train order, departure time and arrival time of the every train at every station. Trains have to operate according to the timetable. However, in actual operation process, trains tend to deviate from planned train timetable due to various uncertain factors, which brings great interference to the operation order and increases the dispatching work. What’s worse, it will lead to a lot of inconvenience for passengers and have a negative impact on railway overall image and market competitiveness. Thus, it is necessary to optimize the structure of train © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 140–147, 2019. https://doi.org/10.1007/978-3-030-04582-1_16

Railway Timetable Diagnostic Analysis Based on Train Operation Data

141

timetable and distribute the redundant time reasonably to make the train timetable with a high degree of flexibility. That is to say, when an unexpected situation occurs, to make sure the train is in a state of driving according to timetable, the trains can quickly adjust to normal. There are 28800 trains running on the railway network each day, the running of the trains can produce huge amounts of performance data every day. These data are valuable and can provide a basis for diagnostic analysis of train timetable quality. By analyzing the train operation record data, we can identify stations and sections that often occur train delays. And we can also excavate the general law of train delay propagation, all of these results can help us adjust the train timetable structure and reduce the occurrence of train delays. Many scholars have studied train delays and have achieved many results. From the research content, the existing research mainly focuses on train delay statistical analysis [1, 2], train delay propagation [3, 4] and train delay reasons analysis [5, 6] and other aspects. From the perspective of research methods, classical statistical models [7, 8] and emerging data mining methods [9] are the most commonly used analytical techniques for trains delay. However, it should be pointed out that most of the existing researches have studied the train delays from the perspective of train dispatching, and there are few studies have studied train delays from the perspective of train timetable generation. This paper mainly studies the train timetable diagnostic analytical framework based on train operation data. The contribution of this paper is two-fold. Firstly, this paper clearly deﬁnes the concept of train deviation from the station and section two levels. Secondly, this paper proposes a train timetable diagnostic analysis framework that consider both planning data and actual data. The remainder of this paper is organized as follows: Sect. 2 gives the related deﬁnition, Sect. 3 proposes the timetable diagnostic analytical framework based on train operation data. Section 4 presents a case study to verify the proposed method, and Sect. 5 offers conclusions.

2 Deﬁnitions and Methods 2.1 atij : atij0 : dtij : dtij0 : dwtij : dwtij0 : rtij : rtij0 : uij : /aij :

Variable and Symbol Planned arrival time of train i at station j. Actual arrival time of train i at station j. Planned departure time of train i from station j. Actual departure time of train i from station j. Planned dwell time of train i at station j. Actual dwell time of train i at station j. Planned running time of the train i between the j 1th and the jth station. Actual running time of the train i between the j 1th and the jth station. Indicates whether train i stops at station j. Arrival delay judgment function of train i at station j.

142

C. Xu et al.

/dij : /j ðrtÞ:

2.2

Departure delay judgment function of train i at station j. Section running time deviation judgment function.

Related Deﬁnition

Deﬁnition 1 (Deviation time): The difference between the actual time and the planned time is called the deviation time. The formal expression of the deﬁnition is as follows: tijd ¼ tij tij0

ð1Þ

n o Where tij 2 atij ; dtij ; dwtij ; rtij , tij0 2 atij0 ; dtij0 ; dwtij0 ; rtij0 , and according to Eq. (1), the deviation time of the arrival time, departure time, dwell time, and running time of the actual timetable relative to the planned timetable can be calculated. Figure 1 shows a schematic diagram of the deviation of train operation, in which the black line represents the planned train path and the red line represents the actual train path.

j −1

rtij

i

i′ dwtij′

j dwtij

j +1 Fig. 1. Schematic diagram of train operation deviation

Deﬁnition 2 (Degree of deviation): The degree of deviation refers to the proportion of the absolute value of the actual time and the planned time to the actual time. Speciﬁc deﬁnition see formula (2). tij tij0 Dij ¼ ð2Þ tij

Railway Timetable Diagnostic Analysis Based on Train Operation Data

143

Dij is a measure of whether the train is driving according to the timetable, and the value is between 0 and 1. The closer the value of Dij is to 0, the closer the train actually operation to the planned operation. It also means that the train timetable is of good quality. Conversely, the closer the value of Dij is to 1, the greater the deviation between the train actual operation and the train planned operation. It also means that the train timetable is of poor quality. Deﬁnition 3 (Train delay): Train delay is one of the train deviations, it is also the train operation index that railway transport companies and passengers are most concerned about. Different countries and regions have different deﬁnitions of train delays. Considering the actual situation of China’s railway operation, this paper gives the following deﬁnition of train delay. If the delay of the train at originating station, destination station or relatively large intermediate station exceeds the allowable threshold, then the train is said to have a delay, otherwise, the train is said to be on time. Namely, for train i, depart from station 1, pass through the station 2; 3; ; m 1, and arrive at station m. If the entire running process of the train satisﬁes Eq. (3), then we can say train i delay occurred. Conversely, if the train’s operation satisﬁes Eq. (4), the train i is said to be running at the punctual point. /di1 þ

X

uij /dij þ

j¼2;;n1

/di1 þ

X

X

uij /aij þ /ain 1

ð3Þ

uij /aij þ /ain ¼ 0

ð4Þ

j¼2;;n1

uij /dij þ

j¼2;;n1

X j¼2;;n1

3 Research Framework 3.1

Deviation-Based Visualization Technology

According to the deﬁnition in Sect. 2, we introduce deviation-based visualization technology. The core of this technology is to divide the degree of deviation is divided into ﬁve levels, and then the chromaticity diagram theory is used to indicate the degree

Table 1. Classiﬁcation for the degree of deviation Dij

Deviation description

0 (0, 0.1] (0.1, 0.3] (0.3, 0.5] (0.5, 1]

Not deviate A slight deviation A moderate deviation A large deviation A serious deviation

144

C. Xu et al.

of deviation by the depth of different colors. The deeper the color, the more serious the deviation. Conversely, the lighter the color, the smaller the deviation. The detailed deviation classiﬁcation is shown in Table 1. 3.2

Train Timetable Diagnostic Analysis Framework

The proposed train timetable diagnostic analysis framework (see Fig. 2) mainly includes three parts: data integration, data analysis and timetable diagnostic. Among them, data integration mainly solve the heterogeneity problem of data from different sources. Data analysis mainly refers to adopt statistical method and data mining technology to analyze train operation data. For one hand, we can conduct comparative analysis of the train section running time and train stop time based on the planned timetable and the actual timetable, to check out the reasonableness of railway technical operation time standards. For another hand, we can use the correlation analysis to analyze the causes of delayed trains. Based on data analysis results and visualization techniques, we can perform train timetable diagnostic analysis. Such as identify most often delay section and station, discover train delay propagation and the causes of delayed trains. All of these can provide evidence for making adjustments of the structure of train timetable or optimizing the layout of the redundant time. Planned timetable

Actual timetable

Data integration Feedback

Descriptive statistical analysis Data analysis

Correlation analysis Deviation-based visualization

Timetable diagnosis

Identify stations, sections that are prone to delays Analyze the relationship between delayed trains

Fig. 2. Train timetable diagnostic analysis framework

4 Case Study This paper takes Puxiong-Xichangbei (a section of the Chengdu-Kunming railway line), which is a single track conventional railway, as the research object to verify the analytical framework presented in this paper. The experimental data includes 11 trains and 19 stations. The analysis results are shown in Table 2, Figs. 3 and 4.

Railway Timetable Diagnostic Analysis Based on Train Operation Data Table 2. Train deviation at each station on a certain day

Correlation coefficient graph

K113 K117

K145

K165 K1501

K9471

K9483 X5633

T8865 T8869

Fig. 3. Correlation analysis of the delayed trains

145

146

C. Xu et al. 50

100

150

-20

0

20

40

60

50 100

0

50 100 150

0

Arriv al.delay

15

25

0

Departure.delay

-20 0 20 40 60

0 5

Dwell.time.dev iation

Running.time.dev iation

0

50

100

150

0

5

10

20

Fig. 4. The correlation between the different delays

In Table 2, red represents a positive deviation, namely train delay. Blue represents a negative deviation, that is, namely train arrives early. The darker the color, the larger the deviation. It can be easily found that most trains will occur delay, and few trains will arrive early. That is because it is not allowed arrive early, except at the terminal station. As can be seen from Table 2, S2, S3, S4, S5, S6, S7, S8, S9, S10, S11, S12, S13, S14, S15, S16, S17 are stations where trains are prone to delays. This is because these stations have lower levels, insufﬁcient line capabilities, and limited scheduling techniques. They are weakly resistant to train delays. We also found that the delay of the cargo train is more serious than the passenger train. In Fig. 3, blue represents a positive correlation, red represents a negative correlation, and the darker the color, the greater the correlation. It can be seen that there is a positive correlation between most of the delayed trains, and only a small delay has a negative correlation between the trains. Moreover, the phenomenon of train delay aggregation is obvious, which indicates that the train delay has a phenomenon of propagation, and the propagation decreases with distance. As can be seen from Fig. 4, there is a strong linear relationship between departure delay and arrival delay. And the relationship between other delays is not obvious.

5 Conclusions Timetable diagnostic analysis has a feedback on train timetable generation, which is an important way to improve the quality of the timetable. This paper deﬁnes the concept of train deviation, train delay, and proposes a timetable diagnostic analysis framework.

Railway Timetable Diagnostic Analysis Based on Train Operation Data

147

The results of case study show the proposed analysis framework can effectively identify stations and sections that are prone to delays and analyze the association between delayed trains. More importantly, our results can visually show the rationality of the train timetable structure. Which can provide a reference for optimizing train timetable structure. The future research direction is to introduce deep excavation technology into the timetable diagnostic analysis process. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702, 2016YFC0802208), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No. 2016X006-D), Service Science and Innovation Key Laboratory of Sichuan Province (KL1701), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Büker, T., Seybold, B.: Stochastic modelling of delay propagation in large net-works. J. Rail Transp. Plann. Manag. 2(1), 34–50 (2012) 2. Yaghini, M., Khoshraftar, M.M., Seyedabadi, M.: Railway passenger train delay prediction via neural network model. J. Adv. Transp. 47(3), 355–368 (2013) 3. Yuan, J., Hansen, I.A.: Optimizing capacity utilization of stations by estimating knock-on train delays. Transp. Res. Part B: Methodol. 41(2), 202–217 (2007) 4. Büker, T., Seybold, B.: Stochastic modelling of delay propagation in large networks. J. Rail Transp. Plann. Manag. 2(1–2), 34–50 (2012) 5. Higgins, A., Kozan, E.: Modeling train delays in urban networks. Transp. Sci. 32, 346–357 (1998) 6. Han, S.H., Yun, S., Kim, H., et al.: Analyzing schedule delay of mega project: lessons learned from Korea train express. IEEE Trans. Eng. Manag. 56(2), 243–256 (2009) 7. Kecman, P., Goverde, R.M.P.: Online data-driven adaptive prediction of train event times. IEEE Trans. Intell. Transp. Syst. 16(1), 465–474 (2015) 8. Goverde, R.M.P.: Punctuality of railway operations and timetable stability analysis, Netherlands TRAIL Research School, The Netherlands (2005) 9. Marković, N., Milinković, S., Tikhonov, K.S., et al.: Analyzing passenger train arrival delays with support vector regression. Transp. Res. Part C: Emerg. Technol. 56, 251–262 (2015)

Research on Train Operation Diagram Compilation of Urban Rail Transit Cross-Line Operation Gong Qing(&) Department of Railway Engineering, Sichuan College of Architectural Technology, Chengdu 610399, Sichuan, China [email protected]

Abstract. With the continuous improvement of urban rail transit networks in major cities, the networking feature of Urban rail transit is becoming increasingly prominent. The actual networking is to organize the train cross-line operating, realizing “interconnection” between different lines and vehicles. In this paper, the train operation diagram of urban rail transit cross-line operation is studied from three aspects: compilation principle, preparation steps and process, and computer programming platform of train operation diagram. Keywords: Urban rail transit Cross-line operation Train operation diagram

1 Introduction According to “the 2017 Urban Rail Transit Industry Statistical Report”, Until the end of 2017, there have been 165 urban rail transit lines opened in 34 cities in China (excluding Hong Kong, Macao and Taiwan), with a total operating length for 5033 km. The number of opening lines and mileages are shown in Table 1 [1, 2]. From the table, It can be known that many cities have transformed from single-line operations to multi-line network operations as the opening of continuous new line, Urban rail transit network has gradually formed. Most of the urban rail transit networks that have been built are networks which are realized by passenger transferring between different network lines, it isn’t truly “networked”. The truly networked is to realize “interconnection” between different lines and vehicles, that is, organize the train cross-line operation. It can improve the utilization rate of Urban rail transit train, make full use of line capabilities, reduce the travelling time of passengers, relieve passenger flow in peak time, especially passenger pressure in the peak hour at transfer station. Train Operation Diagram is comprehensive documentation in urban rail transit production, to prepare cross-line train operation diagram reasonably can ensure Urban rail transit operating safely, efﬁciently and efﬁciently. In this paper, the train operation diagram of urban rail transit cross-line operation is studied from three aspects: compilation principle, preparation steps and process, and computer programming platform of train operation diagram. © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 148–155, 2019. https://doi.org/10.1007/978-3-030-04582-1_17

Research on Train Operation Diagram Compilation

149

Table 1. China urban rail transit lines and mileage in 2017 City

Shanghai Beijing Guangzhou Shenzhen Nanjing Hong Kong Chongqing Wuhan Tianjin Dalian Suzhou Chengdu

Trafﬁc mileage (km) 628.3 591.7 368.9 283.6 348.2 230.80 261.1 237.5 166.0 154.49 121.00 179.4

Number of operation lines 15 21 13 8 9 9 6 7 5 5 4 6

Peak passenger trafﬁc (thousand) 11792 12694.3 9000.0 3943.4 3320.0 –

Average daily passenger trafﬁc (thousand) 9318 10025 6805 3540 2276 –

2400 3024.1 1116.1 – – –

1900 1963 845 348 413 154

2 Compilation Principle of Cross-Line Train Operation Diagram The networking train operation diagram should be prepared by uniﬁed planning, which is advantageous to efﬁcient operation for each line, advantageous to integrity for entire operating network, and advantageous to overall operational efﬁciency and efﬁciency. The following principles should be followed when preparing cross-line train operation diagram: (1) The Transportation Mode and routing path for cross-line passenger should be arranged based on carrying capacity of speciﬁc route, and minimize the number of train crossings, the train path of multiple cross-line trains should be scheduled preferentially [3]. (2) The capacity utilization ratio of adjacent line should be limited to a certain level, it can reduce the influence on other line operation diagrams for late arrival at connection station, and a large amount of buffer time should be reserved to increase the flexibility of train operation diagram when scheduling the train path of cross-line train. (3) Late risk is related to the type of cross-line, the number of times, the capacity utilization ratio of each line, sharing or separate platform at connection station, the layout of the train path, etc. (4) The time range of train access and train handing at cross-line train boundary can be determined based on reasonable arrival and departure time range of trains at departure station and destination, and then the speciﬁc train access and handing time can be determined integrated on line operating conditions, afterward, train

150

G. Qing

path can be drawn adopted forward inference method and backward inference method. The drawing of train path must reflect the pre-set stop schedule plan. (5) The connection of the operation diagram should be considered comprehensively, include technical conditions of the adjacent lines, and transportation organization mode to achieve a seamless connection and facilitate passenger travel. During the preparation process, the rational layout of changing off-line train and in-line train should be focused to ensure reasonable arrival and departure time at departure station, destination, and stations along the route, which can attract passengers and meet the technical working time standards of the trains. (6) Drawing the train path should follow “ﬁrst changing off-line train after in-line train”, while the carrying capacity is tight, the in-line train path can be drawn completely, and then choose suitable train path for cross-line train based on reasonable operation time at the connection station, and ﬁnally determine the inline train path [4].

3 Compilation of Cross-Line Operation Diagram 3.1

Compilation Steps for Cross-Line Operation Diagram

When compiling the train operation diagram for cross-line operation train, ﬁrstly, determining the train operation plan. Secondly, determining the priority of the drawing train, the train whose starting and ending station at lines with tight capabilities should be determined preferentially. Finally, drawing in-line train path after determining the layout of cross-line train path, appropriate adjustments can be made for cross-line train path during the process of drawing, while overall train operation plan should be coordinated after adjustment. 3.2

Compilation Process of Cross-Line Operation Diagram

The networking train operation diagram should be prepared by uniﬁed planning, in order to improve transportation efﬁciency and management, reduce the level of dispatching command management, and strengthen the concentration of transport organizations. The compilation process of cross-line operation diagram is as shown in Fig. 1. (1) Prepare cross-line operation plans for each line in a uniﬁed way When preparing operation plan, the influence to each line generated by the crossline plan should be considered comprehensively, the planners should have experience in the preparation of relevant line operation plans and have a deep understanding of the operation plans for each line. Or operation plans are prepared by the planners drawing the cross-line train operation plan for relevant lines. (2) Prepare in-line train operation plan based on a deﬁned cross-line operation plan Usually the cross-line operation plan shouldn’t be changed after it is determined. However, if the cross-line operation plans have a greater impact on the in-line operation when in-line operation planes are prepared, the ﬁrst step should be recompleted to re-determine the cross-line operation plans.

Research on Train Operation Diagram Compilation

151

Compiling Organization for operation diagram

Planner for line1

Change and feedback

cross-line operation dia-

Planner for cross-line operation diagram

Planner for line2

operation diagram

Planner for line N

Reply Reviewer

Fig. 1. The compilation process of cross-line operation diagram

(3) Check the completed overall operational plan When overall operational plan is completed, There should be a specialist to check it, after the approval is passed, the plan is completed. Otherwise, return to the ﬁrst step or the second step according to the results. 3.3

Adjustment Process of Cross-Line Operation Plan

When the trains in the line have a large area delay, emergency command team should be established to consider the beneﬁts of the overall network comprehensively, In case the adjustment of each line is mainly based on the beneﬁt of the line while adjustment measures implemented on its own, this will lead to a lower efﬁciency for entire network operation adjustment. (1) When the external interference is not very serious, the dispatching system can automatically or the dispatcher manually adjust the cross-line train to a certain extent, the following steps can be referenced (Fig. 2): ① Adjust the in-line operation plan; ② Notify the relevant line dispatching command system that the crossline plan has changed; ③ The relevant line planners adopt the change of the cross-line train plan and adjust the in-line plan accordingly. Or the relevant line planners believe that the adjustment of the cross-line train is not appropriate and refuse to adopt it. The relevant line planners shall negotiate and resolve it, and when the negotiation is inconsistent, it shall be submitted to the higher authorities for resolution. (2) When it is necessary to increase or cancel cross-line operation trains temporarily, the following steps can be referenced (Fig. 3):

G. Qing

line1

line2

adjust the in-line operation plan

Line N

Fig. 2. The adjustment process of situation (1)

total dispatcher

increase or cancel cross-line operation trains temporarily

152

line1

adjust the in-line operation plan

adjust the in-line line2 operation plan adjust the in-line operation plan

line N

Fig. 3. The adjustment process of situation (2)

① The higher authorities (total dispatchers) formulate cross-line train plans for temporarily increased cross-line train, or designate temporarily cancelled cross-line trains; ② The relevant line increase or cancel the corresponding cross-line train according to the instructions of the higher authorities, and adjust the inline train operation plan and the corresponding cross-line train plan accordingly; ③ If adjusting the in-line plan affects other cross-line trains (try to do not adjust other cross-line train plans as a principle), process (1) can be referenced.

Research on Train Operation Diagram Compilation

153

(3) When external interference may cause serious delay or has caused a large area delay, the following steps can be referenced (Fig. 4): ① Establish an emergency command team; ② According to the influence range of delay, formulate an adjustment plan for cross-line trains based on the overall network operation and passenger retention; ③ According to the adjustment plan for cross-line train, formulate the train adjustment plan for corresponding line; ④ Each line implements a corresponding adjustment plan.

adjust the cross-line operation plan

emergency command team

line1

adjust the in-line operation plan

adjust the in-line operation plan

line2

line N

adjust the in-line operation plan Fig. 4. The adjustment process of situation (3)

4 Computer Programming Platform of Cross-Line Train Operation Diagram Cross-line operation diagram compilation platform refers to establish cross-line operation compilation platform through computer means, the system can compile the train operation plan for the entire network, and store the completed operation diagram information into system data platform, and present the compilation process and compilation results to the user by the way of graphs and charts. 4.1

Basic Requirements of Computer Programming Platform for CrossLine Train Operation Diagram

The following requirements should be considered when compiling the cross-line train operation diagram in computer programming platform: (1) Flexible operation diagram compilation (2) Complete veriﬁcation analysis function (3) Friendly graphical human-machine editing interface [5, 6].

154

4.2

G. Qing

System Functional Framework Design of Computer Programming Platform for Cross-Line Train Operation Diagram

System functional framework design of computer programming platform for cross-line train operation diagram is as shown in Fig. 5. computer programming platform for cross-line train operation diagram

maintenance and management of Basic data

preparation and adjustment of train operation diagram

preparation and adjustment of The train using plan

indicator calculation and analysis of operation diagram

calculation and Analysis of line carrying capacity

drawing train operation diagram and Generation of statistics report

query and humancomputer interaction processing

interface and integration processing

decision support management

Fig. 5. System functional framework design of platform

(1) Maintenance and management of basic data It can maintain and manage the basic data of entire network, verify the basic data reported by each line uniformly, so that the basic data of entire network is consistent and can be updated in time. It can achieve data entry, addition, deletion, query, statistics, update, backup etc., and generate history ﬁles, statistical ﬁles, interface ﬁles, report ﬁles of operation diagram compilation [7]. (2) Preparation and adjustment of train operation diagram Line network center compiles the train operation diagram of the entire network based on passenger flow plans from related business management department, departure interval, number of trains, stop plan at different periods, etc. Adjustment function adjusts and modiﬁes train operation diagram according to adjustment range and demand. The results of the compilation and adjustment are stored in the database and downloaded to the ATS Operations Subsystem of each line, supplied to train trafﬁc control. (3) Preparation and adjustment of the train using plan The line network center considers the train using plan among lines and within lines as a whole, compiles and adjusts the train using plan. (4) Indicator calculation and analysis of operation diagram The system can calculate indicator of operation diagram, and establish an indicator analysis and evaluation system to evaluate the generated train operation diagram, and provide decision basis for managers. (5) calculation and analysis of line carrying capacity The system can use the solving model of the carrying capacity, calculate the and generate line carrying capacity, and analyze and evaluate on this basis to search an optimization strategy for comprehensive utilization of capabilities, supplied decision basis for trains number and mode. (6) Drawing train operation diagram and Generation of statistics report

Research on Train Operation Diagram Compilation

155

The system can draw and generate the corresponding train operation diagram, timetable and train routing diagram, and calculate and generate relevant indicator tables and capability analysis tables for the operation diagram. (7) Query and human-computer interaction processing The system can provide relevant technical parameters and information of the train operation diagram for the entire network, and provide means for inquiring, supervising and controlling the compilation process (8) Interface and integration processing The system can achieve data transferring, human-computer interaction, software operation, conversion and output of shared information, and reception and conversion of external information. (9) Decision support management The system can establish a decision support management module, and use the management decision model and management method to carry out the system’s auxiliary decision management.

5 Conclusion Compared with the closed single line operation, the cross-line operation organization has a higher complexity. The operation organization of each line transforms from independent operation to highly coupled cross-line operations because of the existence of cross-line operation trains. So, preparing a scientiﬁc and reasonable cross-line train operation diagram can ensure the operation of entire network efﬁcient, safe and reliable. Acknowledgments. This research was supported by the Fundamental Research Funds for the Central Universities (2682017CX018, 2682017CX022), the National Key R&D Program of China (2017YFB1200702, 2016YFC0802208), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D).

References 1. Urban Rail Transit 2017 annual statistics and analysis report. Urban Rail Transit (04), 6–25 (2018) 2. Wang, Y., Zhao, Y., Pang, J.: Statistics and analysis of China’s urban rail transit operation lines in 2017——the ﬁfth of China urban rail transit annual report. Urban Rail Transit Res. 21 (01), 1–6 (2018) 3. Chen, H.: Research on linkage of transportation between passenger’s dedicated lines and existed railways. Beijing Jiaotong University (2008) 4. Qu, M.: Research on passenger transportation mode under high-speed railway network. Southwest Jiaotong University (2014) 5. Wang, K., Ni, S.: Survey on research and application of train operation diagram making system. J. Transp. Eng. Inf. 14(03), 75–82 (2016) 6. Ni, S.: The train working diagram making system of China railway. Southwest Jiaotong University (2013) 7. Ni, S., Lv, H., Li, H.: Study on the train working diagram making system. Railw. Transp. Econ. (07), 32–35 (2001)

Research on the Optimization for the Utilization of Passenger Train Stock of Existing Railway Zhiqiang Tian(&), Zhenbo Yang, Tianyi Sheng, and Rui Zhang School of Trafﬁc and Transportation, Lanzhou Jiaotong University, Lanzhou 730070, China [email protected], [email protected], [email protected], [email protected]

Abstract. The effective utilization of passenger train stock is the key to realize railway transportation plan and reduce the cost of operation. This paper researches on the utilization of passenger train stock of existing railway and proposes a set partitioning model, which is constrained by the train stock class and the carrying capacity of different train pairs based on the train operation plan. By limiting the number of connected train pairs, this problem is transformed into maximum matching problem of general graphs and solved by Lingo12. The optimal result shows that the optimization method proposed in this paper can do the pre-optimizing work before making the train working diagram and cut down the amount and cost of train stocks efﬁciently. Keywords: Passenger train

Train stock Set partitioning model

1 Introduction For the high acquisition cost of passenger train stock, it is meaningful for rail’s operation to optimize the effectiveness of train stock’s utilization which is aimed at use less train stocks complete the speciﬁed passenger train lines. From the perspective of both theoretical research and practical application, the utilization of passenger train stock mainly has two modes which called ﬁxed application and recycled application. The former mode is using certain train stocks to ﬁx certain passage train lines, while the later mode has a view at a series of train lines that adjacent lines can share a same train stock. Therefore, the recycled application is usually more economy on the residence time at stations than ﬁxed application. Since the departure and arriving time of all passenger trains is unknown before the compilation of train working diagram, the common use of recycled application is to link adjacent train pairs or train lines based on the completed train working diagrams. For this disadvantage, the relationship among linked train lines may not the optimal. To improve the utilization efﬁciency of passenger train stock, many scholars have done a lot of research recently. Liu and Sun [1, 2] established an optimized model and algorithm for recycled utilization of passenger train at a center based on the research on the effect of operation mode of train set. Xie, Zeng and Xu [3] proposed a modal based © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 156–163, 2019. https://doi.org/10.1007/978-3-030-04582-1_18

Research on the Optimization for the Utilization of Passenger Train Stock

157

on a situation which is under the limit of train sets’ count aimed at the maximum of transportation capacity. Zhu and Li [5] aimed at the minimum wasted residence time at station for departure and destination-arrived train and proposed a model for complete diagram to improve the recycled relationships. Zhang [7], Shi [8], Arianna [9] studied on the recycled relationships for normal passenger train and motor train set based on the reality and gave certain optimization suggestions respectively. Because of the complexity of the problem, applying the research results into application still have to face great challenge. As a matter of fact, the structure of the train working diagram is limited after the compilation work, which leads to a fatal defect that the optimization space is relatively small for the optimization of passenger train stocks. Therefore, we think it is feasible and ideal that train working diagram makers should pre-combine the recycled relationships of train pairs (or train lines) during the period of timetabling.

2 Analysis of Recycled Relationships for Train Sets This study is based on the assumption that it is an imprecision of running time and residence time under the train working diagram’s making period. The estimated travel time is inferred from empirical values or the timetable from pre-making documents. The estimated turnover time Ti for a certain train pairs Li can be calculated as follow. Ti ¼ tis þ tix þ tib þ tiw

ð1Þ

Where tis and tix are estimated travel time for upward and downward, respectively (min); tib and tiw are the standard value of the overhaul time at own depot and tun-back depot, respectively (min). When the passenger train pairs Li (contains an upward train and a downward train) uses its own train stock, the number of train stocks ni could be calculated as ni ¼ dTi =1440e. For the characteristic of this, the more Ti =1440 closer to an integer, the less additional residence time wasted at the stations of both sides. In practical, the standard value of overhaul time is relatively ﬁxed, while the running time for upward and downward usually vary in a large range. So there usually have waste of efﬁciency because it is hard to make all train pairs turnover time approach k times of 1440 min. It has been proved by the application experience that recycled mode is a useful measure to enhance the efﬁciency in passenger train stock operation. Nevertheless, the complexity of existing railway passenger trains leads to the obstacle on application of this mode for the capacity, speed and class among all kinds of normal passenger trains. Unlike the high-speed railway lines, the trains travelling time on the existing railway lines are difﬁcult to be guaranteed, and the chain effect of delay may bring severe global delay which leads to the global recycled relationship is worse than local recycled relationship in existing railway lines. To weigh the advantage and disadvantage, local recycled relationships can be rather feasible than the global recycled relationship because it can greatly enhance the efﬁciency of some train paths and avoid the risk of delay for the relatively low rate of punctuality.

158

Z. Tian et al.

3 Optimization Model Based on the Train Operation Plan 3.1

Problem Description

In this paper, a train pairs Li is a basic unit for recycled relationships which includes the train class, carrying capacity and all kinds of standard values of operation time. Li can be deﬁned as Li ¼ ðtis ; tix ; tib ; tiw ; Ti ; ni ; Ri ; Qi Þ, where Ri is the class of this train pairs, Qi is the carrying capacity of this train pairs. The scheme of recycled train pairs is combined by certain train pairs which means these train pairs use a group of train stocks together. The scheme also contains these parameters: train class, carrying capacity and turnover time. This can be factored as Pj ¼ ðTj ; nj ; Rj ; Qj ; Sj Þ, where Pj refers to a feasible scheme, Tj is the turnover time of m P this scheme (min), Tj ¼ aij Ti ; aij is a binary parameter that equals 1 if Pj contains Ti , i¼1 l m Tj and 0 otherwise; nj is the required train stocks for this scheme, nj ¼ 1440 ; Sj is the number of train pairs contained in this scheme. Since the difference of train class and carrying capacity among all train pairs, the recycled application must ensure adequate service level and capability. Therefor, for adjusting the difference of train class and carrying capacity, we choose the higher rank and bigger carrying capacity to ensure the feasibility of the recycled scheme, which means the following constraints should be satisﬁed. Rj ¼ maxfaij Ri g

i ¼ 1; . . .; m;

ð2Þ

Qj ¼ maxfaij Qi g

i ¼ 1; . . .; m;

ð3Þ

To relieve the cycled constraint of all trains pairs that use the same train stocks, we could set a parameter K which is the celling for the number of connected train pairs, which means Sj K. For a recycled scheme, we call it a favorable scheme since the cost is lower compared with the sum of all train pairs use the train stock independently. Thus, the essence of the optimization is to search for a best combination of all favorable schemes to approach the minimize cost. 3.2

The Further Optimization Based on the Favorable Schemes

Recycled application is aimed at saving the count of train stocks which devotes to considerable operation cost. The optimal solution of this problem can be obtained by solving a set partitioning problem consist of all favorable schemes which obtained under the constraint of K. The set partitioning model (M1) which is aimed at minimum acquisition cost of passenger train stock can be presented as follow: min Z ¼

n X j¼1

Xj nj Qj PRj P0

ð4Þ

Research on the Optimization for the Utilization of Passenger Train Stock

8 n i > > > . > > > < minð^vi ðtÞ2 ð2Dxi ðtÞÞ ; aacc max ai1 ðtÞÞ ¼ > minðDxi ðtÞ aacc max =xm ; Dxi ðtÞ aacc max > > > > > maxðDxi ðtÞ aabreak max =xm ; aabreak max ai1 ðtÞÞ > > : 0

^vi ðtÞ [ 0; Dxi ðtÞ\0 ^vi ðtÞ\0; Dxi ðtÞ [ 0

ð4Þ

^vi ðtÞ 0; Dxi ðtÞ\0 ^vi ðtÞ 0; Dxi ðtÞ [ 0 ^vi ðtÞ ¼ 0; Dxi ðtÞ ¼ 0

Dxi ðtÞ which can be derived from formula (5) is the distance between the location of train i and the location Dðvi ; vi1 Þ in rear of train i 1. In Fig. 1, the actual headway which negative is between the control train and the front train is greater than the ideal headway Dðvi ; vi1 Þ. The actual headway between the two trains ðxi1 ðtÞ xi ðtÞÞ is ðDðvi ; vi1 Þ Dxi ðtÞÞ. Dxi ðtÞ ¼ xi ðtÞ ðxi1 ðtÞ Dðvi ; vi1 ÞÞ

ð5Þ

The ideal distance Dðvi ; vi1 Þ can be derived from formula (6), where d0 denotes the safety distance between the front end of the control train and the rear end of the front train after they stopping in the case of service braking of the control train under the condition of emergency braking of the front train. The safety distance d0 can be calculated according to d0 = (response time to braking + processing and transmission delay of signal) speed of control train safety coefﬁcient, with the safety coefﬁcient in the range of 1–2. Dðvi ; vi1 Þ ¼ d0 þ L þ Sci ðvi ðtÞÞ Sui1 ðvi1 ðtÞÞ

ð6Þ

The acceleration of train i is adjusted according to the acceleration of train i 1, and thus the running status of train i is changed. The control acceleration of train i is given by formula (7). ai ¼ Dai þ ai1

ð7Þ

4 Study of a Simulation Case The effectiveness of the MAS based coordinated control model is analyzed by designing a speciﬁc scenario. The simulation model is built with the designed scenario simulating by making use of Simulink of Matlab. The initial speed of the leader is

Coordinated Control Method of Virtually Coupled Train Formation

231

200 km/h; the speeds of other trains in the train group are in the range of 198–216 km/h (Fig. 3). Initial state 181km/h

3010m

2929m

x4

Leader

187km/h

208km/h

2827m

x3

200km/h

206km/h

x2

2634m

x1

x0

Fig. 3. Designed situation of simulation case

The train group reaches the stable coordinated status at a simulation time of 700 s, at the same time, the actual headway between the control train and the front train gradually reached the ideal distance of 181 m. In this process, the initial headways are quite long while the actual speeds of all the trains except the leader increasing substantially at the early stage and then reducing gradually to 200 km/h. The train group can reach stable status when these trains traveled at the same speed as the leader (Fig. 4). Actual distance(m) 3000

2000

Train speed(km/h) 350 Train1 Train 2 Train 3 Train 4

Leader Train1 Train 2 Train 3 Train 4

300 250

1000

200 0

0

500

1000 1500 2000 2500 Simulation time (s)

3000

150

0

500

1000 1500 2000 Simulation time(s)

2500

3000

Fig. 4. Relationship among simulation time, actual headway and train speed

The scenario is simulated by MAS based VC, speed based moving block and distance based moving block, as shown in Fig. 5. The results show that the train control method of distance based moving block provides the lowest efﬁciency, while the methods of VC and speed based moving block achieving substantial efﬁciency improvement.

232

L. Liu et al. Actual headway(m) 3000 2000 1000 0

0

200 MAS based VC

400

600 Simulation time(s)

800

Speed based moving block

1000

1200

Distance based moving block

Fig. 5. Relationship of simulation time and headway in different train control methods

MAS based VC which is proposed in this paper, achieves an average headway of the control trains of 194 mm when speed based moving block and distance based moving block is 263 m and 2152 m respectively.

5 Research Conclusions This paper applies MAS to the control of VCT formation running in the section between stations. A MAS based coordinated control model is established accordingly. The proposed model achieves real-time dynamic adjustment, and can ensuring the safe and efﬁcient operation of train groups. Through simulation analysis, it turns out to be feasible and efﬁcient to apply the MAS based coordinated control method to the VCT formation control. In future research, it is necessary to develope an integrated technology combining MAS based coordinated control method with dispatching organization. Acknowledgments. This work was supported by the “National Key R&D Program of China” (2017YFB1200700).

References 1. Bock, U., Varchmin, J.U.: Erhöhung der Streckenauslastung durch virtuelle Zugverbände. Tagung der VDI-Gesellschaft Fahrzeug- und Verkehrstechnik, pp. 315–324. VDI Verlag GmbH, Duesseldorf (1999) 2. Jia, L., Qin, Y., Wang, L.: Scientiﬁc and technological innovation of rail transportationtrends and tasks. J. Beijing Jiaotong Univ. (2016) 3. Bock, U., Bikker, G.: Design and development of a future freight train concept – “virtually coupled train formations”. IFAC Proc. Volumes 33(9), 395–400 (2000) 4. Braun, I., König, S., Schnieder, E.: Multi agent systems for rail transport. Pravni život Časopis za pravnu teoriju i praksu, 14 (2005) 5. Goikoetxea, J.: Roadmap towards the wireless virtual coupling of trains. In: International Workshop on Communication Technologies for Vehicles, Communication Technologies for Vehicles. Lecture Notes in Computer Science, pp. 3–9. Springer International Publishing, Cham (2016)

Coordinated Control Method of Virtually Coupled Train Formation

233

6. Schumann, T.: Increase of capacity on the Shinkansen high-speed line using virtual coupling. Int. J. Transp. Dev. Integr. 4(1), 666–676 (2017) 7. Wang, R., Dong, X., Li, Q., Ren, Z.: Distributed adaptive time-varying formation for multiagent systems with general high-order linear time-invariant dynamics. J. Franklin Inst. 353 (10), 2290–2304 (2016) 8. Qu, C., Cao, X., Zhang, Z.: Multi-agent system formation integrating virtual leaders into artiﬁcial potentials. J. Harbin Inst. Technol. 46(5), 1–5 (2014) 9. Pan, D., Zheng, Y., Zhang, C.: On intelligent automatic train control of railway moving automatic block systems based on multi-agent systems. In: Proceedings of the 29th Chinese Control Conference, Beijing, China, pp. 4471–4476. IEEE (2010) 10. Xun, J.: On cooperative train control based on agent and cellular automation. Beijing Jiaotong University (2011) 11. Xue, X., Pan, J.: An overview on evolutionary algorithm based ontology matching. J. Inf. Hiding Multimedia Sig. Process. 9(1), 75–88 (2018) 12. Wang, K., Mondal, S.K., Chan, K., Xie, X.: A Review of Contemporary E-voting: Requirements, Technology, Systems and Usability. Data Sci. Pattern Recogn. 1(1), 31–47 (2017)

Coordination Evaluation Index of High-Speed Railway Network Capacity Tao Chen1,2,3, Jie-ru Zhang1,2,3, Hong-xia Lv1,2,3(&), and Jin-shan Pan1,2,3 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected], [email protected], [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, Sichuan, China 3 National United Engineering Laboratory of Integrated and Intelligent Transportation, Chengdu 610031, Sichuan, China

Abstract. Make efﬁcient use of the transportation capacity of high-speed railway network has very important signiﬁcance. This paper studies the connotation of high-speed railway network capacity coordination ﬁrstly. Then, coordination evaluation index system of high-speed railway network capacity is constructed into two aspects: the coordination of transport demand and network capacity, the coordination of point and line of high-speed railway network. At the same time, eight indexes are built and are all given the calculation method. At last, actual high-speed railway data are ensured to verify the validity of these indicators and provide ideas for optimizing the structure of high-speed railway network and improving its capacity utilization. Keywords: High-speed railway

Network capacity Coordination index

1 Introduction High-speed railway construction is an important way to expand the effective supply and improve the quality of railway service. How to make efﬁcient use of the transportation capacity of high-speed railway network, and meet the rapid growth of passenger flow demand is a problem worthy of in-depth study. The coordinated evaluation of high speed railway network capacity is an important means to identify the bottleneck of high-speed railway network capacity and optimize the utilization of high-speed railway

This research was supported by National Natural Science Foundation of China (Project No. 61703351), the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018), the National Key R&D Program of China (2017YFB1200702,2016YFC0802208), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), and Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017-RK0000378-ZF). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 234–241, 2019. https://doi.org/10.1007/978-3-030-04582-1_28

Coordination Evaluation Index of High-Speed Railway Network Capacity

235

network. At present, the research on the evaluation method of railway network capacity is mainly limited to the coordination of station and line system, but the overall coordination research of high-speed network system is lacking [1–3]. This paper ﬁrst deﬁnes the connotation of capacity coordination of high-speed railway network. Then this paper constructs an index system of capacity coordination evaluation of high-speed railway network from the aspects of trafﬁc volume, structure and utilization of railway network trafﬁc structure. And the actual high-speed railway data are used for veriﬁcation.

2 High-Speed Railway Network Capacity Coordination Connotation Analysis Coordination of system refers to the harmonious coexistence of its constituent systems under the action of self-organization within the system and regulatory management activities from outside (i.e., other organizations) in order to achieve the optimization of the overall goals of the system [4]. High-speed railway network capacity coordination can be made out from the following two aspects: • High - speed rail network capacity and transport demand coordination The high-speed railway network is mainly to provide transportation services for passengers. The ultimate goal of the entire high-speed railway network system is to achieve the coordination of transportation demand and capacity. The study on the coordination of transportation demands and capacity of the high-speed railway network will help the railway transportation department build a reasonable transportation scale and network structure layout of the high-speed railway network according to the passenger transportation needs of various regions, and maximize the advantages of high-speed railway network operation. Because the transportation demand of the highspeed railway network has a very obvious imbalance on time and space, the coordination between the capacity of the high-speed railway network and the transportation demand requires not only coordination in quantity, but also coordination in time and space distribution. • High-speed railway point and line capacity coordination High-speed railway point and line capacity coordination is the internal coordination of high - speed railway network system. Point system passenger contains hub station, central station, EMU (Electric Multiple Unit) depot. Line system contains railroad section, rail connecting line ect. The research on the coordination of the network capacity of high-speed railway can determine whether the internal transportation equipment and facilities of the network meet the actual transportation demand, at the same time, it can help the transportation department to make adjustment plans for the weak links in the system and gives full play to the transportation capacity of various equipment.

236

T. Chen et al.

3 Construction of Coordination Evaluation Index System of High-Speed Railway Network Capacity According to the connotation of high-speed railway network capacity coordination, the evaluation index of high- speed railway network capacity coordination is shown in Fig. 1: High - speed railway network capacity coordination evaluation index system

Target Layer

Criterion Layer

Index Layer

High-peed railway point and line capacity coordination

High-speed rail network transportation demand and capacity coordination

High speed railway network transport demand satisfaction

Daily seat

occupanc y rate of high-speed railway

High - speed railway network capacity utilization equilibrium

Highspeed railway network congestion

High speed railway station and line sections capacity matching degree

High speed railway station and line sections capacity utilization matching degree

EMU depot and connecting lines capacity matching degree

EMU depot and connecting lines capacity utilization matching degree

Fig. 1. Coordination evaluation index system of high-speed railway network capacity

① High - speed railway network transport demand satisfaction The transport demand satisfaction rate is used to reflect the extent to which passenger demand is served in the railway transport network. It can be expressed in terms of the ratio of the number of passengers transported by the railway network to the transportation demand. The train operation scheme provides passenger transport capacity and passenger flow demand satisfaction degree. The closer this index gets to 1, the better the index is. This index is calculated as follows: P s a¼P s

qs q0s

ð1Þ

In this formula, qs is total passenger flow from high-speed railway between OD pair s, q0s is passenger demand potential between OD pairs. ② Daily seat occupancy rate of high-speed railway network The index refers to the ratio of passenger load on the high-speed railway network station line to train trafﬁc. This index is helpful to understand whether the passenger demand of high-speed railway is in harmony with the transportation capacity of high

Coordination Evaluation Index of High-Speed Railway Network Capacity

237

speed railway network in spatial distribution. The closer this index gets to 1, the better the index is. This index is calculated as follows: bij ¼

qj bi pj B

ð2Þ

In this formula, qj is daily passenger volume of each station or section of the highspeed railway network,ten thousand people/day. bi is passenger volume fluctuation coefﬁcient, different coefﬁcient can represent the passenger volume of different periods. pj is the number of trains per day in each station or section of the high-speed rail network, train/day. B is the average passenger capacity of high-speed rail network trains, person/train. ③ High - speed railway network capacity utilization equilibrium The utilization equalization of the capacity of the high-speed railway network in each period refers to the variance of the capacity utilization rate and average capacity utilization rate of each station in each period. The smaller the index is, the more balanced the network capacity is in space, otherwise it is less balanced. This index is calculated as follows: 1 ni 1 X ni 2 ci1 ¼ ð Þ ð3Þ M Ni M i2M Ni In this formula, ni is the daily trafﬁc of each station or section of the high-speed railway network, Ni is daily capacity of each station or section of the high-speed rail network, M is the total number of stations and lines. In order to facilitate coordination judgment, we use formula (4) to calculate: ci2 ¼ 1

1 ni 1 X ni 2 ð Þ M Ni M i2M Ni

ð4Þ

④ High-speed rail network congestion High railway network congestion refers to the proportion of the total number of stations in the railway network whose capacity load exceeds a certain limit. It directly reflects the average distribution level of railways with high capacity load [5]. Generally, when the capacity utilization exceeds 80%, it is considered to be crowded. In order to judge the coordination degree, we use the statistics of non-congested sections. The larger the index, the better, indicating no congestion. This index is calculated as follows: The unit step function is introduced in the formula, that is, for each section or station in each time period, there are the following relations: ( IðiÞ ¼

1; Nnii 0:8 0; Nnii \0:8

ð5Þ

238

T. Chen et al.

di ¼ 1

1X IðiÞ M i2M

ð6Þ

⑤ High - speed railway station and line sections capacity matching degree The transit capacity of HSR station and line interval refers to the passing capacity of station and line interval under the current transportation organization condition according to the actual demand. The index is the ratio of the sum of the passing capacity of the adjacent sections of a station to the passing capacity of the station. It can reflect the coordination between the station and the line capacity of the network in the operation process. The closer the index is to 1, the better. This index is calculated as follows: P i Nsection j2D ei ¼ ð7Þ i Nstation i In this formula, Nsection is the section carrying capacity of one line which next to the i station, train/day. Nstation is the carrying capacity of one station, train/day.

⑥ High - speed railway station and line sections capacity utilization matching degree This index refers to the ratio of the capacity utilization rate of each line in each station of the high-speed railway network. Formula (2) is used to calculate the capacity utilization rate of each line in each station. It can reflect the coordination between the station and the line capacity utilization of the network in the operation process. The closer the index is to 1, the better. This index is calculated as follows: fi ¼

bisection bstation

ð8Þ

In this formula, bisection is the carrying capacity utility ratio of one line which next to the station. bstation is the carrying capacity utility ratio of one station. ⑦ EMU depot and connecting lines capacity matching degree The design and maintenance capacity of EMU depot refers to the maintenance capacity that the EMU can achieve under the current transportation organization condition in daily operation. Contact line carrying capacity is the capacity that can be reached under the current transportation organization conditions. This index refers to the ratio between the carrying capacity of the connecting line and the maintenance capacity of the EMU depot, which can reflect the coordination between the two capabilities in the operation process. The closer the index is to 1, the better. This index is calculated as follows: gi ¼

i Nsection i Ndepot

ð9Þ

Coordination Evaluation Index of High-Speed Railway Network Capacity

239

i In this formula, Nsection is the daily carrying capacity of connecting line, train/day. i Ndepot is daily maintenance capacity of EMU depot, train/day.

⑧ EMU depot and connecting lines capacity utilization matching degree This index refers to the ratio between EMU depot maintenance capacity utilization and connecting line carrying capacity utilization. It can reflect the coordination of the two capabilities in the operation process, and the closer the index is to 1, the better. This index is calculated as follows: fi ¼

bisection bdepot

ð10Þ

In the formula, bisection is the daily carrying capacity utilization of connecting line, bidepot is EMU depot maintenance capacity utilization. It needs to be explained that there are many ways to calculate the capacity of station line section and the maintenance capacity of EMU depot, one of which can be used to calculate. Since the intermediate station is not a capacity bottleneck, the ⑤, ⑥ index only calculates the matching degree between the hub station and the line sections.

4 Example Take the relevant data of a certain high-speed railway as an example to evaluate its capacity coordination. The railway is 505 km long, with a design speed of 350 km/h and an initial operating speed of 300 km/h. It use CRH2C EMU, 3 min tracking interval, 4 min arriving interval, 8 h maintenance time. The maximum daily design capacity of line section is 240 pairs/day. Due to the existence of different speed trains, the actual daily maximum capacity of the line is 150 pairs/day. This line has 11 stations, and the stop time of interval station is less than 4 min. The starting and terminal stations carrying capacity is calculated according to the application plan of the arrival line. Station data and capacity utilization along the high-speed railway are shown in Table 1. In addition, the whole line consists of two EMU depots which connected with station A1 and A11 separately, their maintenance capacity are as follows: EMU depot B1 can repair 40 EMU per day, and 64 pairs trains in and out of depot per day, EMU depot B2 can repair 40 EMU per day, and 64 pairs trains in and out of depot per day. 8 vehicles are marshalled into an EMU, 610 passengers per EMU. The high-speed rail line capacity and capacity utilization are shown in Table 2. According to the above data, the capacity utilization index of the high-speed railway is calculated as the last column of Table 1 and the last two columns of Table 2. The remaining indicators are shown in Table 3.

240

T. Chen et al. Table 1. Station data and capacity utilization along the high-speed railway

No.

Station

1 2 3 4 5 6 7 8 9 10 11

A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11

Number of arrival and departure lines 12 4 4 6 4 4 4 6 4 8 15

Actual maximum carrying capacity (pairs/day) 263 136 136 240 136 136 136 240 136 240 433

Number of trains received (train/day) 145 64 64 64 64 64 64 64 64 62 100 (contains 34 empty train)

Daily capacity utilization 0.65 0.55 0.47 0.47 0.27 0.47 0.47 0.47 0.27 0.26 0.23

Table 2. High-speed rail line capacity and capacity utilization Section

1 2 3 4 5 6 7 8 10 11 12 13

B1 EMU depot -A1 A1–A2 A2–A3 A3–A4 A4–A5 A5–A6 A6–A7 A7–A8 A8–A9 A9–A10 A10–A11 A11–B2 EMU depot

Train number (train/day) 26

Daily passenger volume (person/day) 0

Daily capacity utilization 0.42

Daily seat occupancy

79 31 32 32 32 32 32 31 31 33 17

14052

0.53 0.21 0.22 0.22 0.22 0.22 0.22 0.21 0.21 0.22 0.27

0.29 0.74 0.71 0.62 0.61 0.47 0.45 0.50 0.50 0.47 0

13851 12063 11808 9193 8738 9484 9559 0

0

According to the coordination degree standard in reference [3] (Table 4), the indexes show: from the aspect of demand-capacity coordination, this high-speed rail transport capacity can completely meet the demand of transportation, but seat occupancy is discordant; from the aspect of point-line coordination, station and line sections capacity and capacity utilization are matched well, but EMU depot and connecting lines capacity and capacity utilization are matched moderately.

Coordination Evaluation Index of High-Speed Railway Network Capacity

241

Table 3. Coordination indexes of high-speed railway High speed railway network transport demand satisfaction

Daily seat occupancy rate of high-speed railway network

High speed railway network capacity utilization equilibrium

Daily highspeed railway network congestion

High - speed railway station and line sections capacity matching degree

High - speed railway station and line sections capacity utilization matching degree

EMU depot and connecting lines capacity matching degree

EMU depot and connecting lines capacity utilization matching degree

0.93

0.54

0.984

1

A1

A11

A1

A11

B1

B2

B1

B2

0.81

0.49

0.96

0.95

0.58

0.55

0.65

0.68

Table 4. Coordination degree standard Coordination degree

M, Num=1, jump to Step 2. Repeat Step 2-Step 4 Until Stopping criteria met Obtain a series of global optimization solutions

Fig. 1. A general framework of the tabu search algorithm

We bring in a parameter Yd;awt ¼ ðfawt ; yd;awt ; zan0 ;pdðnÞ ; ud;a;av Þ to indicate demand d 2 D transportation planning. If yd;awt [ 0, ud;a;av are known, xd;awt , wd;a;av are also welldetermined as per the formula (14) and (16), respectively. If fawt [ 0 and yd;awt [ 0, thus zan0 ;pdðnÞ ¼ 1,ud;a;av ¼ 0 as per the formula (11), (13)–(16); in contrast, if fawt [ 0 and yd;awt ¼ 0, thus zan0 ;pdðnÞ ¼ 0 or 1,ud;a;av 0. Perturbation is performed by train capacity utilization threshold and demands on virtual arcs. (a) For links a, if fawt [ 1, adjust the value of fawt with fawt [ 0. For demands d 2 D, if delivered by links a, maintain zan0 ;pdðnÞ ¼ 1, and adjust the value of yd;awt ; if no delivered by links a, adjust the value of zan0 ;pdðnÞ ,yd;awt . (b) For links a, adjust train departure time to period t0 , if fawt [ 0, with fawt ¼ 0 and fawt0 [ 0. The car flow classiﬁcation adjustment is similar to (a). (c) For links a, if fawt ¼ 0 8t 2 T, increase the value of fawt with fawt [ 0. The car flow classiﬁcation adjustment is similar to (a).

5 Model Testing In this section, a 14-yard railway network as shown in Fig. 2, is used to test the model and solution approach. The scheduled duration is set to 12 h, with 12 periods. 1

3

2 4

7

5

8

9

6

12 10 13

11 14

Fig. 2. The structure schematic diagram of railway network

Freight Railroad Service Network Design Problem

257

Table 1 shows the technical operating parameters of each yard are listed. The expenditure time passing through the segment and the segment capacity and shipment demands are not list due to conﬁned length. Others parameters is set as follows: h ¼ 50; ma ¼ 50; a ¼ 3 000; Fa ¼ 500; Fav ¼ 200. Table 1. Relevant station parameters Yards Rn;t Bn;t Tn;ct da

Yards Rn;t Bn;t Tn;ct da

1 2 3 4 5 6 7

8 9 10 11 12 13 14

100 160 80 150 80 160 170

8 7 6 6 6 7 7

3 3 4 4 4 3 4

9.2 8.4 9.8 10.0 9.3 10.2 9.6

150 130 140 160 170 130 120

8 6 7 8 7 7 8

4 4 3 4 4 4 5

9.7 9.2 9.3 9.8 9.5 10.0 9.5

As shown in Table 2, when the problem size is large, the TS algorithm can effectively obtain a better optimization solution in the same time, compared to CPLEX.

Table 2. The performance analysis of tabu search algorithm Problem size Yards Demands 10 14 14

11 60 100

CPLEX Solution time (second) 164 7200 7200

Objective value 125000 _ _

Tabu search Solution time (second) 48 525 689

Objective value 125000 640000 1005000

6 Conclusion In this paper, we proposed a dynamic FRSNDP model by means of space-time network, considering the flexible inbound time of shipment demands, which bridges the gap between routing shipments and scheduling trains. Experimental results show that the proposed model and TS algorithm is feasible, promote to enhance the reliability of tactical operation planning, and provide market-sensitive service to customers. Acknowledgments. This research was supported by the National Key R&D Program of China (2016YFC0802208), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), and Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), and Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017-RK00-00378-ZF).

258

X. Liu et al.

References 1. Crainic, T.G., Fweland, J.A., Rousseau, J.-M.: A tactical planning model for rail freight transportation. Transp. Sci. 18(2), 165–184 (1984) 2. Keaton, M.H.: Designing railroad operating plans: a dual adjustment method for implementing Lagrangian relaxation. Transp. Sci. 26(4), 263–279 (1992) 3. Huntley, C.L., Brown, D.E., Sappington, D.E., et al.: Freight routing and scheduling at CSX transportation. Interfaces 25(3), 58–71 (1995) 4. Li, H.Y.: Railway freight transportation organization mode reform and relevant technologies research. Beijing Jiaotong University (2008) 5. Haghani, A.E.: Formulation and solution of a combined train routing and makeup, and empty car distribution model. Transp. Res. Part B: Methodol. 23(6), 433–452 (1989) 6. Kwon, O.K., Martland, C.D., Sussman, J.M.: Routing and scheduling temporal and heterogeneous freight car trafﬁc on railnetworks. Transp. Res. Part E: Logist. Transp. Rev. 34(2), 101–115 (1998) 7. Jha, K.C., Ahuja, R.K., Sahin, G.: New approaches for solving the block-to-train assignment problem. Networks 51(1), 48–62 (2008) 8. Zhu, E., Crainic, T.G., Gendreau, M.: Scheduled service network design for freight rail transportation. Oper. Res. 62(2), 383–400 (2014) 9. Lin, B.L., et al.: Optimizing the freight train connection service network of a large-scale rail system. Transp. Res. Part B: Methodol. 46(5), 649–667 (2012)

Research on Emergency Rescue Plan for Cross Region Comprehensive Transportation Network Shan Huang1,2,3, Jie Zhang1,2,3(&), Hui Zhang1,2,3, and Changyu Liao1,2,3 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China

Abstract. Cross-regional comprehensive transportation network needs a perfect emergency rescue plan system to ensure accurate and efﬁcient emergency rescue for various emergencies. This paper studies the construction of the emergency plan system for the cross-regional comprehensive transportation network and the content of the emergency rescue plan. Based on the two-tier planning model, the site selection of the emergency resource supply point of the cross-regional comprehensive transportation network is studied, and then the emergency resource supply is provided. The layout of the points, assuming that each mode of transportation has the shortest path, and selecting the multiple transportation combination modes of the cross-regional comprehensive transportation network through the emergency resource scheduling model aiming at minimizing the time cost, and determining the scheduling plan of the emergency resources, provide basis and technical support for emergency rescue of crossregional comprehensive transportation network. Keywords: Emergency rescue plan Cross-region Comprehensive transportation network

1 Concept and Connotation When an emergency occurs, the emergency rescue plan can guarantee the accurate implementation of emergency rescue work in the ﬁrst time and take emergency rescue measures timely. The emergency rescue plan can clarify the scope of work, responsibility requirements, emergency rescue procedures and other precautions at all levels. According to the type of emergency and the degree of hazard, a series of detailed emergency plans are provided for the responsibility of emergency rescue personnel, the selection of emergency rescue vehicles, the allocation of emergency resources, and the emergency rescue action program. © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 259–267, 2019. https://doi.org/10.1007/978-3-030-04582-1_31

260

S. Huang et al.

The region is a broad concept, and human production and life are within the scope of a certain region. A spatial unit with some homogeneity and aggregation is usually referred to as a region [1]. In the event of a disaster accident in various regions, it is easy to cause a chain reaction, resulting in a series of secondary and derivative disasters, causing regional damage, and various transportation lines may also suffer different degrees of damage. At the same time, the storage of materials in each region is limited. Once a large amount of relief materials is needed, local materials cannot meet the demand. Cross-regional rescue will effectively improve the current situation of emergency rescue that cannot be met by emergency rescue in the existing area. At the same time, the comprehensive transportation network will effectively combine various modes of transportation, use different means of transportation and road network for emergency rescue, and improve the efﬁciency of the rescue. The cross-regional comprehensive transportation network emergency rescue plan is the key guiding basis for emergency response in the new era. It can control and use the comprehensive transportation network for event warning and transmission of information in the ﬁrst time after an emergency. Emergency response measures such as emergency response, plan decision-making, order issuance, and transportation to cross-regional emergency rescue resources are carried out to protect the lives and property of the people. Cross-regional comprehensive transportation network emergency rescue should consider two situations: (1) Comprehensive transportation network emergency rescue: Various transportation modes cooperate with each other to carry out emergency rescue materials transportation, and ensure that sufﬁcient emergency relief materials are transported to the affected areas in the shortest time. From the research of comprehensive emergency rescue plans for roads, railways, aviation and water transport, the cooperation of various modes of transportation will highlight their respective advantages, save the transportation time of rescue equipment and materials, and improve the efﬁciency of emergency rescue. (2) Cross-regional emergency rescue: In the past, the scope of emergency rescue was generally limited to one region or inter-region. However, in combination with recent domestic and international emergencies, it can be seen that in the actual emergency rescue process, the cooperation and solidarity of the multiple regions can complete the rescue work. The cross-regional comprehensive transportation network emergency rescue plan is based on the analysis and evaluation of the severity of emergencies, comprehensive analysis of basic data information such as emergency rescue action program, resource layout, and various transportation combinations. At present, the research on the emergency rescue plan for the cross-regional comprehensive transportation network mainly includes the establishment of the emergency rescue plan system, the research on the theory and technology of the emergency rescue plan, the basic structure of the emergency plan, the content, the preparation of the emergency plan, the preparation method and so on.

Research on Emergency Rescue Plan

261

2 Research Status at Home and Abroad (1) Foreign countries have not yet given a complete comprehensive emergency response plan for comprehensive transportation network. Therefore, the research status of the corresponding emergency system is mainly introduced from the railway, aviation and highway in the comprehensive transportation network. (1) Railway: The Japanese railway emergency plan was prepared by the Japanese railway department in cooperation with public security ﬁreﬁghting and medical institutions, and was revised and improved by the Ministry of Land, Infrastructure, Transport and Tourism for ﬁnal determination. The content covers: emergency training, guiding principles of the exercise, standard implementation process of the emergency rescue plan, grading response and different scale standards for speciﬁc events, emergency facilities, resource security and related technologies. The French railway system prepares the emergency rescue plan on a provincial basis, and is responsible for taking the lead in the provinces and regions. The states and the National Infrastructure Management Agency work together to discuss a reasonable plan. The content covers: the functions and tasks of emergency management personnel, the principle of communication to the media, the corresponding plan management procedures, the composition of the emergency rescue department, and resource security. Deutsche Bahn AG is responsible for the preparation and revision of the railway system emergency plan, and the federal government cooperates with the emergency department to ensure that the plan is effective. The content covers: the tasks of the railway organization, the contractual provisions of the company and the government, the ability and condition of the responsible person, the emergency management response procedure, the communication mechanism with the media, and the classiﬁcation of emergencies [2]. (2) Highway: Japan has a good road emergency response mechanism, and rapid medical linkage rescue can use emergency rescue time effectively; the United States has a comprehensive highway emergency plan system, standardized processes can shorten emergency response time; Japan and Germany have advanced the road trafﬁc information center, the timely delivery of emergency rescue information is guaranteed by advanced technology. (3) In aviation, as early as 1974, the United States formulated the relevant aviation emergency rescue bill. In order to prevent the failure of the commercial communication system, a full-service Boeing 747 aircraft was used as the communication command center as a major natural disaster response command platform to ensure the real-time dynamics of rescue information. The Ministry of Emergency Situations is the center of the Russian aviation emergency management system. It adopts vertical form of emergency management, introduces talents and builds a large-scale professional team to ensure professional rescue forces. The Japanese military’s aviation selfdefense team is responsible for aviation emergency rescue work. The maritime security department is responsible for assisting emergency rescue; in the

262

S. Huang et al.

case of severe disasters, when more rescue equipment is needed, civilian helicopters will also carry out emergency materials transportation under the recruitment of the Japanese government. In 1970, Osaka Fireﬁghting Helicopter Rescue Assistance Construction was established. It is one of the earliest teams in the world for aviation emergency rescue. The emergency rescue system in Europe has been very sound. The national emergency command center has been established 10 years ago [3].

(2) Domestic emergency rescue research and discussion will also be carried out from the perspectives of railway transportation, air transportation, water transportation and road transportation. Water transportation includes marine transportation and inland transportation. Ocean transportation is not considered on the outside of the land. The transportation time required for inland transportation is relatively long, and it is not superior to the tasks that railway and road transportation can undertake in China’s emergency rescue work. It mainly studies the emergency rescue system of domestic railways, highways and aviation. (1) In terms of railways, the railway department has issued various railway emergency plans in accordance with the requirements of the State Council on the preparation of emergency plans at all levels since 2005. At present, the railway system has built a three-level emergency plan system, which consists of three levels consisting of the China Railway Corporation, various railway bureaus, and stations, and conducts emergency rescue command to issue emergency rescue plans [4]. The railway emergency plan plays an extremely important role in dealing with railway emergencies. However, there are still some shortcomings in the railway emergency plan: the division of decentralized management is not detailed, the query is difﬁcult, and there is no sharing. At the same time, not only the railway emergency plan has problems, but the connectivity plans between various transportation networks are weak, and cross-regional emergency rescue research is still insufﬁcient. Therefore, under the comprehensive transportation conditions combined with various transportation modes, the research on the emergency plan of the comprehensive transportation network across regions is extremely important. (2) On the highway side, according to the 2006 National Emergency Response Plan for Public Emergencies, most areas have established a joint emergency rescue system for road trafﬁc accidents, but they are still in the stage of research and improvement [5]. The proposed vertical and horizontal linkage mechanism for emergency rescue of road trafﬁc emergencies can promote the cooperation of cross-regional and inter-departmental emergency joint rescue work. (3) In aviation, the advantage of aviation emergency rescue is to save emergency time and reduce rescue links. However, the high cost of aviation emergency rescue and the scarcity of professional rescuers have become a problem in aviation emergency rescue. China’s aviation emergency rescue system is a new topic and is currently under planning. The main aviation emergency rescue forces include the Military, the Ministry of Communications, the Salvage and

Research on Emergency Rescue Plan

263

Salvage Bureau, the Civil Aviation Administration of China, the Police Aviation and the Ministry of Civil Affairs. The civil aviation emergency rescue plan system is comprehensively improved. The existing plans include the Overall Plan for Emergency of the Civil Aviation Administration of China and the State Emergency Plan for Disposal of Civil Aircraft Flight Accidents.

3 The Construction of Emergency Rescue Plan System Cross-regional comprehensive transportation network emergency rescue requires regional governments to incorporate ﬁnancial, public security, ﬁre protection, ﬁrst-aid, trafﬁc police, public utilities, health, armed police and other departments, related personnel, materials, rescue vehicles and other resources into a system in accordance with emergency laws and regulations. In the middle, establish an emergency command center and a corresponding emergency organization management structure, and rationally arrange and dispatch emergency resources in each region. Rely on the emergency rescue information system for daily monitoring and prevention, take appropriate emergency rescue command according to different levels of emergencies, and coordinate the joint actions of various departments and rescue resources in a timely manner. Cross-regional comprehensive network emergency rescue is a complex system project that requires careful and detailed preparation. A comprehensive emergency response plan can ensure the most effective control measures in the ﬁrst time of an emergency, thereby reducing personal danger and property losses. The cross-regional comprehensive transportation network emergency rescue plan system is constructed from the following 8 points: (1) (2) (3) (4) (5) (6) (7) (8)

Top-level design of the plan. Analysis of the needs of emergency rescue plans. Emergency plan management team. Analysis of emergency hazard analysis with emergency rescue plan capability. Preparation of the plan. Review and release of the plan. Pre-planned training drills. Evaluation and revision of the plan.

4 Cross-Regional Comprehensive Transportation Network Emergency Rescue Plan Content The basic contents of the inter-regional comprehensive transportation network emergency rescue plan are as follows: (1) General, mainly describes the purpose of preparation, preparation basis, working principle and scope of application of the plan. (2) Identiﬁcation of hazard sources, covering the investigation and understanding of hazard sources, analyzing the degree of hazard, and discovering hazard events that are likely to be triggered and their scope and adverse effects afterwards.

264

S. Huang et al.

(3) Organization and responsibilities, covering the details of various emergency organizations and responsibilities, and the organizational framework of the organization. All kinds of emergency agencies are composed of leading responsibility agencies, ofﬁces, local institutions and expert groups. Input, analysis, and output of each department’s tasks, the process of connecting tasks and report results, and the facilities and equipment required to complete the tasks. (4) Prevention and early warning, covering the determination of the warning level and the transmission of information after the warning. (5) Emergency response, this element is the most critical one in the plan, which can be decomposed into hierarchical response, emergency response action, information sharing and processing, emergency communication (information reporting and processing), command and coordination, emergency response, Ambulance medical care, personnel safety protection, resource allocation, event investigation and analysis, consequence assessment, press release, etc. (6) Post-disposal: After the emergency rescue phase is completed, the repair and after-care work after the emergency rescue is particularly important. From the sudden emergency to the normal state, the time, personnel, and form status change, so according to different types. The emergency should be preestablished with a corresponding recovery plan. It includes post-processing measures, quotations and insurance, lessons learned and revision suggestions, as well as recovery work arrangements after emergency rescue, and safety recovery of emergency systems. After the restoration, the recovery management checklist should be ﬁlled to check the work of each personnel, the use of each equipment, and the operation of each transportation network. (7) Safeguard measures: Covering safeguards in communications, emergency equipment, rescue teams, transportation vehicles, medical and health facilities, public security, security technology, and material reserves. When an emergency occurs, timely organize emergency medical rescue and on-site sanitation disposal at the incident, and coordinate the professional rescue team to send experts and relevant emergency rescue equipment for rescue. On-site emergency rescue personnel should wear protective equipment as needed, and evacuate, transfer, and resettle the masses in a safe area under the premise of ensuring their own safety, and protect them. (8) Emergency preparedness: construction, training and drills of the emergency rescue professional team. (9) Supplementary rules, covering plan management and revision, special awards and fault liability. (10) Attachments, covering emergency rescue work flow chart, emergency personnel joint method, risk assessment and emergency capability review results.

Research on Emergency Rescue Plan

265

5 Key Technologies for Emergency Rescue of Cross-Regional Comprehensive Transportation Network 1. Method of preparing the plan Emergency rescue plan apply scenario - task - ability to prepare technical methods for emergency preparedness. The traditional emergency rescue plan focuses more on rescue than on preventive monitoring. There is no prior prediction and discussion on the hazards caused by unexpected emergencies, and the preparation method cannot reflect the preparation process of the core content of the emergency plan. Scenario analysis is a method to study the formation of corresponding plans in combination with the location layout of emergency resource warehouses under the assumption of certain conditions. Scenario analysis can ensure the rationality and operability of the crossregional emergency rescue plan preparation, that is, from the emergency scenario setting, the response strategy formulation, and the emergency capability guarantee, etc., according to the steps to improve the emergency rescue plan. The response procedure is the corresponding emergency action for the set scenario. The scenario setting summarizes the experience by reviewing past event cases and formulates response measures based on the hypothetical scenario to achieve a response to the rescue goal. 2. Compilation techniques When an emergency occurs, the emergency monitoring facility equipment monitors the environmental changes for the ﬁrst time, identiﬁes the hazard source, transmits the information to the emergency command system through the monitoring platform, and conducts an emergency warning of the comprehensive transportation network through the early warning system, information data, search for the type of emergency rescue plan that meets the requirements. The emergency commander combines the emergency decision response system with the auxiliary decision of the emergency rescue response system to carry out emergency rescue operations. The task is assigned by the chief commander to each emergency support department, the emergency rescue team leading the emergency management team at each level, and the on-site emergency rescue team to carry out rescue according to the contents of the emergency plan. The aviation, water transport, railway and highway departments actively carry out the transportation of rescue equipment, rescue vehicles, relief supplies and rescue personnel, and start the emergency rescue dispatch system to determine the number of rescue personnel and the number of emergency vehicles required in combination with the type and severity of emergencies. The connection node of the emergency material transportation route and the emergency rescue tool ensures that the required resources will be transported to the emergency rescue site for emergency rescue at the ﬁrst time. (1) Location layout of emergency resource supply points The problem of the emergency resource supply point location model needs to be solved: properly locate the provincial and municipal warehouses, scientiﬁcally determine the address, and conﬁgure the quantity of the two-level warehouse material reserve to form a reserve network to ensure the emergency resource guarantee rate. Under the constraint conditions, the cost of total material storage is minimized.

266

S. Huang et al.

According to the classiﬁcation of the emergency rescue plan for the cross-regional comprehensive transportation network, the emergency incidents with less harm are provided by the municipal reserve warehouse for materials. For serious emergencies, the provinces and municipalities jointly provide relief supplies. The ﬁrst phase of the rescue provides emergency resources from the nearest municipal-level relief material warehouse and the nearest provincial-level warehouse to ensure emergency response time [6]. In the second stage, emergency resources are generally guaranteed by provincial material warehouses. The quantity of materials stored in different levels of material storage warehouses in provinces and municipalities can meet the amount of materials required for the respective response level. (2) Selection and scheduling of transportation combination methods After the occurrence of an emergency, combining with the risk analysis to choose transportation means for the comprehensive transportation network: which kind of transportation combination of railway, highway, aviation and water transportation is used for emergency resource transportation. The transportation mode is selected by combining the emergency resource scheduling model with multiple resources and multiple supply points [7]. The time cost of emergency resource scheduling is based on the shortest transportation time between the supply point and the disaster area, and the time cost is converted into transportation cost [8].Combined with the freight cost of the scheduled transportation mode, the optimal transportation combination method is analyzed to minimize the total transportation cost comprehensively. An emergency dispatch plan can determine that a certain amount of emergency resources are transported from a storage warehouse to the disaster site through a particular mode of transport. (3) Determination of the number of various transport vehicles How many transport vehicles are used in the emergency rescue process for material dispatch. If the emergency materials are light goods, the number of transport vehicles is determined in terms of volume, and the heavy goods are determined by the weight as the standard.

6 Summary Emergency rescue measures can be taken in time for emergency rescue to ensure that the emergency rescue work must be carried out in an orderly and accurate manner in the ﬁrst time. The cross-regional comprehensive transportation network emergency rescue plan studies the need for a regional or local area to meet the needs of emergency rescue. Inter-regional rescue will effectively improve the current situation of emergency rescue that cannot be met by emergency rescue in the existing area. It will effectively combine various modes of transportation, take advantage of different transportation tools, road networks for emergency rescue, and improve the efﬁciency of emergency rescue. The cross-regional comprehensive transportation network emergency rescue plan is the key guiding basis for emergency response in the new era. It can control and use the integrated transportation network for event warning and transmission of

Research on Emergency Rescue Plan

267

information in the ﬁrst time after an emergency. Emergency response measures such as emergency response, plan decision-making, order issuance, and transportation to crossregional emergency rescue resources are carried out to protect the lives and property of the people. Acknowledgement. This research was supported by the National Key R&D Program of China (2016YFC0802208), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149,2017RZ0007, 2017015,2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Chen, G., Zhang, X., Jin, Q.: Regional Emergency Management Practices – Preplans, Exercises and Performance. Beijing Chemical Industry Press, Beijing (2008) 2. Lu J.: Research on the preparation method of emergency plan for railway dangerous goods transportation accident. Southwest Jiaotong University, pp. 30–32 (2014) 3. Yu, G.: Aviation Emergency Rescue. Beijing Aviation Industry Press, Beijing (2009) 4. Zhou H.: Pre-plan management and resource allocation optimization in railway emergency management. Beijing Jiaotong University, p. 61 (2011) 5. Liu, J.: Research on the emergency rescue linkage mechanism of highway trafﬁc emergency. People’s Communications Publishing Co., Ltd. (2017) 6. Mete, H.O., Zabinsky, Z.B.: Stochastic optimization of medical supply location and distribution in disaster management. Int. J. Prod. Econ. 129, 76–84 (2010) 7. Wang, Y., Jin, F.: Research on improved evolutionary programming algorithm for emergency resource scheduling problem. Oper. Res. Manag. 21(4), 29–33 (2012) 8. Chen, W.: Research on emergency location and scheduling method for multiple resources and multiple supply points. Xi’an University of Electronic Science and Technology, pp. 27–28 (2014)

Selection Conditions Analysis of Passenger Transport Mode for Fast Passenger Network Qiangfeng Zhang1,2,3,4, Shaoquan Ni1,2,3(&), Dong Chen1, and Wentao Li1 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 School of Information Science and Technology, Southwest Jiaotong University, Chengdu 610031, China

Abstract. The design and selection of the passenger flow mode should be as close as possible to the transportation enterprise’s revenue and passenger demand. Due to the transportation resources and considering the transportation efﬁciency, it is difﬁcult for the transportation enterprise to meet all the passengers’ travel time, fare and service frequency. This paper discusses the components of the generalized cost of passenger travel, and studies the relationship between the fare rate, departure frequency, transit time, transfer convenience and other parameters of the direct and transit passenger flow modes, and analyzes the passenger’s voluntary choice of transfer satisfaction condition. Based on the background of fast passenger transportation network, the principle and method of dividing regional subnets in fast passenger transportation network and the characteristics and selection principles of intermediate conversion multi-nodes are given. The work done can provide reference for the design of passenger train service attributes, and serve as the basis for transportation companies to guide passengers to mode selection according to their own resources. Keywords: Express traveler network Generalized travel cost

Passenger flows transportation mode

The selection and design of the passenger flow mode is the basis and premise of the train development plan, which mainly includes two modes: direct and transit. With the large-scale construction of China’s high-speed railways, as of the end of 2017, China’s high-speed railways have a total mileage of more than 25,000 km, and have formed a rapid and complex passenger transportation network with high-speed railways, intercity railways and passenger dedicated lines. On a road network with wider coverage and larger scale, scientiﬁcally and reasonably determining the passenger flow mode and improving the overall efﬁciency of the fast passenger transport network have always been the hot spots of the transportation production department and scholars. At present, the relevant results of the domestic passenger flow mode have their own characteristics © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 268–279, 2019. https://doi.org/10.1007/978-3-030-04582-1_32

Selection Conditions Analysis of Passenger Transport Mode

269

in research ideas; goals and content. Li et al. [1] and Zhang [2] mainly discuss the selection of passenger flow mode based on qualitative perspective and give more macro recommendations. Such research results have a certain guiding role for the actual transportation production organization. In comparison, more research is based on quantitative methods and has achieved richer research results. The content and angle of the research can be summarized as follows: (1) Establish a selection parameter relationship model, study the passenger flow mode selection from the perspective of enterprises or passengers, or give the proportion of direct and transit trains in the passenger flow mode. Xiang [3] established a model selection parameter relationship model based on passenger travel expenses, and discussed the selection conditions of passenger transfer mode; Yan et al. [4] from the perspective of transportation enterprises, the problem of train operation mode is determined by the collaborative optimization of the train running section and the train category. The method of determining the mode of passenger trains with the transport utility as the core is proposed. Wang et al. [5] analyze the passenger transport mode of railway passengers in China. Based on the preferences of the passengers, the organizational conditions of the direct mode and the medium conversion multiplication mode are discussed, and speciﬁc implementation suggestions are proposed. (2) Discussing and analyzing the method of determining the passenger flow mode from the perspective of methodology [4]; (3) taking a speciﬁc railway as the research object, combining the characteristics of the line, developing the transport mode research, and obtaining a more speciﬁc Suggestions. Zhang [6] proposed that the Beijing-Shanghai high-speed railway should be suitable for the train-based mode of transfer-oriented, and the design principles for the organization of the Beijing-Shanghai high-speed railway, and the speciﬁc design scheme for the passenger flow organization. China’s road network is large in scale, and the whole road network is related to the number of ODs. The passenger flow mode selection is a large-scale combinatorial optimization problem. At the same time, due to the lack of a persuasive evaluation system, it is difﬁcult to select optimization schemes among many feasible schemes, which makes the design schemes have a great effect in practical application. However, it is difﬁcult to consider the network characteristics of the road network by simply studying the passenger flow mode of a certain railway. Based on reasonable travel time, the road network is divided into several regional subnets. Based on this, the passenger flow mode is discussed, and the passengers’ willingness to travel is effectively combined with the transportation resources provided by railway transportation enterprises. The selection and design of passenger flow mode is adopted. It has a good guiding signiﬁcance, and this idea is rarely seen in existing related research.

1 Passenger Flow Mode The passenger flow mode is also called the train travel mode. When the passenger flow is transmitted to any OD in the road network, the direct or medium transfer mode is adopted. Broadly speaking, the problem of passenger flow mode selection can be understood as the problem of coordinating passenger demand and transportation resources, and coordinating the relationship between passenger travel expenses and

270

Q. Zhang et al.

transportation enterprise efﬁciency. With the continuous improvement of the coverage of China’s fast passenger transportation network, it is of great practical signiﬁcance to carry out research on passenger flow mode with fast passenger transportation network as the object. This paper will focus on the passenger flow mode of China’s railway rapid passenger transport network. Unless otherwise stated, “cross-line” passenger flow refers to the passenger flow in each line of the fast passenger transport network, excluding the cross-line of the ordinary speed line to the high speed line. The nonstop or inter-change mode has certain advantages and disadvantages for transportation enterprises and passengers. Analysis of the advantages and disadvantages of the two and the adaptation conditions are beneﬁcial to determine the selection conditions of the passenger flow mode and the key factors affecting the passenger mode selection. The speciﬁc analysis is shown in Table 1.

Table 1. Analysis of advantages, disadvantages and adaptation conditions of direct and medium conversion multipliers Project Advantage

Nonstop mode (1) Improve passenger travel convenience, reduce travel fatigue and enhance travel experience (2) Save travel time to some extent

Disadvantage

(1) Low frequency of travel, less travel opportunities (2) The setting of the “rectangular” sunroof in the ﬁxed period brings trouble to the opening

Adaptation conditions and considerations

(1) The right journey (2) Certain passenger flow requirements (3) Line technical conditions, number of multiple units and maintenance conditions are allowed

Transfer mode (1) The frequency of driving is high, and the transfer will bring more rides, that is, the service frequency is higher (2) Simpliﬁed line plan (3) Improve the ability to pass the busy section (1) Increase the difﬁculty of station operation organization (2) Increase the capacity of the station to pass (3) Increase passenger travel fatigue (1) Long distance not suitable for direct trains (2) Transfer station capacity and passenger organization ability can meet the demand

In the inter-change mode, the multiple units has a short turnaround time and high utilization efﬁciency, which can reduce the vehicle bottom demand of the multiple units, and at the same time, can effectively improve the utilization rate of the section according to the passenger flow. There is less interference in train operation between different lines, which reduces the difﬁculty of formulating transportation plans and command and dispatch. The train runs at a high frequency and has a high punctuality rate, which can effectively improve the reliability of the transfer. In the direct mode, passengers have high travel convenience, but the EMU turnaround time increases, increasing the demand for multiple units.

Selection Conditions Analysis of Passenger Transport Mode

271

2 Regional Subnets and Passenger Nodes 2.1

Division of Regional Subnets

From the experience of foreign drives, in addition to the widespread use of transfer in Japan, European countries mainly use nonstop mode and reasonable transfer as the driving principle [4]. At present, China’s multiple units trains mainly adopt point-topoint operation mode, and the number of direct trains accounts for a relatively high proportion. Although this mode of travel meets the needs of passengers to reach their destinations, it reduces the passenger transfer and is more conducive to attracting passengers. At the same time, it also leads to the tension of some trunk segments, which restricts the comprehensive capability of the entire high-speed rail transport network. It should be noted that the land area of European countries is generally small, and the train journey under the direct mode is relatively small. The long waiting time for transfer and the baggage handling during the transfer, the extra burden of trafﬁc in the city, and the uncontrollable factors in the transfer are important reasons why most passengers are unwilling to choose to transfer. For a long time, China’s railways have also tended to reach the direct mode when formulating the existing passenger trains. Even the two OD nodes of the long journey, as long as their political and economic status, passenger flow and other requirements meet certain requirements, direct trains are opened. China’s fast passenger transportation network with high-speed railway as the main body has the characteristics of fast running speed, high driving frequency, high degree of modernization of passenger stations, and more scientiﬁc passenger transport organization. As the scale of the road network continues to expand, it is also possible to transfer between the same station and even the same station platform, which greatly reduces the cost of transfer. From the passenger’s point of view, according to the results of ergonomic research, the general travel 6 h will cause physical fatigue, so 8 h should be taken as the upper limit of the passenger’s single travel distance [7]. According to the train operation map implemented on July 1, 2015, there are 135 pairs/d of multiple units trains with train running time of more than 8 h in China. From the analysis of ticket sales data, most of the passengers on these trains are medium and short-distance passengers. The average number of passengers is 297 people/train, which is less than 1/3 of the quota [7]. Due to the use of ﬁxed-time rectangular maintenance sunroofs, the high-speed railway has also objectively increased the difﬁculty of direct trains for long journeys. Therefore, except for some cities, due to their political and economic reasons, direct access to direct trains should be limited to a certain range. This paper deﬁnes this range as the passenger flow transport area subnet. The passenger flow transport area subnet is divided according to the following principles: (1) The diameter of the regional subnet should be controlled within the travel time range of 6–8 h, and at least one central passenger flow node and several passenger flow nodes should be included; (2) The central passenger flow node should have strong motor reserve and maintenance capability, and the geometric center of the area cannot be directly determined as the central passenger flow node.

272

Q. Zhang et al.

(3) Traverse all selected central passenger flow nodes, and divide the regional subnet according to the diameter range described in (1) until all the passenger flow nodes are included in at least one regional subnet. According to the division method of the above-mentioned area subnet, the division result of the area subnet is not unique, that is, there are multiple division results on the same road network. Based on the regional subnet, when discussing the passenger flow mode between any two OD nodes in the subnet, it is necessary to further clarify which passenger nodes on the path through which the OD pair is suitable as the intermediate transfer multiplier, which involves the division of passenger flow node and the selection of the inter-change node. 2.2

Passenger Flow Node Division

Referring to the classiﬁcation criteria of the existing passenger node classiﬁcation, the stations in China’s fast passenger transportation network (in the case of no confusion, the station name is represented by the city name) are graded [8]. The main basis for the division includes: (1) the average daily passenger volume sent in recent years; (2) the number of trains currently originated; (3) the social attributes of the nodes; (4) the status and function of the nodes in the road network; (5) multiple units maintenance productivity layout; (6) Medium and long-term road network planning for railways. Class I: Road network passenger transport centers, generally located in megacities such as Beijing, Shanghai, Guangzhou, Wuhan, Chengdu, Xi’an, etc. Class II: Regional Passenger Transport Center is a regional integrated passenger transport hub in the railway network, generally a provincial capital and some important big cities such as Harbin, Shenyang, Jinan, Zhengzhou, Nanchang, Fuzhou, Kunming, Nanning, Lanzhou and Urumqi. Wait. Stations (cities) where multiple lines are concentrated in the railway express passenger transportation network should also be included in this level, such as Xuzhou and Zhuzhou. Class III: Stations with originating and ﬁnal capabilities, but not belonging to Classes I and II, generally ground-level stations, such as Guangyuan and Mianyang. Class IV: General stations that do not have initial or ﬁnal capabilities, generally county-level stations, such as Shuangliu and Guanghan. Class V: Over the line Station is only set up to increase the passing capacity of a certain section. It only handles technical operations such as passenger trains and voyages, and does not handle passenger transportation such as passengers’ boarding and landing. The class I and II passenger flow nodes are generally selected as the inter-change nodes.

Selection Conditions Analysis of Passenger Transport Mode

273

3 Analysis of Passenger Flow Transport Mode Selection Conditions Different generalized travel expenses in nonstop or inter-change mode are important factors affecting passenger selection. By analyzing the composition of passengers’ general travel expenses, the influence of each component on generalized travel expenses can be further clariﬁed, so as to use the adjustment of passenger service product attributes to guide passengers to choose the purpose of passenger flow mode. 3.1

Assumptions

The calculation of the passenger’s general travel expenses and the choice of passenger flow mode are based on the following assumptions: (1) The study of passenger flow mode is based on each passenger flow node in the regional subnet; (2) Multiple transfers will greatly increase the fatigue of passengers. Two or more transfers are not accepted in most cases, so only one transfer is considered. (3) Assume that passengers can transit as scheduled. The transfer time and the extra fatigue cost caused by the transfer are included in the transfer cost, regardless of the transit time and related expenses caused by poor transfer. 3.2

Generalized Cost Calculation for Passenger Travel

Deﬁning the general cost of passenger travel consists of four parts: fare, travel time cost, service frequency, and transfer cost. The fare is multiplied by the fare rate and travel distance, and the joint fare for the transfer plan is considered. Travel time refers to the sum of the time spent by passengers on the train at the starting point of the journey, excluding the waiting time for the transfer; the service frequency is expressed by the average interval between departures within one day, which reflects the passenger’s choice of travel time; the cost is mainly composed of the time and feeling that the passenger spends in the transfer, and both are quantiﬁed as time. Multiply the above three times by the unit time value of the target passenger to obtain the time value cost. The parameter assumes: The two segments that make up the transfer are L1 , L2 ; The corresponding train levels are u1 , u2 ; the average travel time of trains running on L1 and L2 roads is t1 , t2 hours respectively; the corresponding fare rates for L1 and L2 road segments are r1 , r2 , respectively, then the fares corresponding to the road sections are Cp1 ¼ L1 r1 , Cp2 ¼ L2 r2 , respectively, the total fare for the journey is Cp ¼ Cp1 þ Cp2 ; the discount rate of the joint ticket is A; the average time value of passengers is HR yuan/hour, the average transit time is th hours. In addition to the time spent, the additional fatigue time cost for transit to passengers is tew , this parameter indicates the convenience of the conversion multiplication node. For the convenience of calculation, the sum th þ tew of the average transit time and the additional fatigue feeling equivalent time cost is referred to as the transfer time cost; the direct train frequency is fz times/day; under the transfer scheme, the L1 train frequency is fh1 times/day, and the L2 train frequency is fh2 times/day. According to the high-speed rail

274

Q. Zhang et al.

operation experience that has been put into operation, the reasonable arrival time of the train is 6:00–24:00. Based on this, the average departure time interval and time cost consumption are calculated. Direct access method for general travel expenses Cz ¼ Cp þ ðt1 þ t2 Þ HR þ

18 HR fz

ð1Þ

The conversion multiplication method generalized travel expenses Ch ¼ Cp A þ ðt1 þ t2 þ th þ tew Þ HR þ

18 HR minðfh1 ; fh2 Þ

ð2Þ

Only when the generalized travel cost in the transfer mode is lower than or equal to the direct mode, the passenger will actively select the transfer mode, that is, the condition Cz Ch is satisﬁed. Substituting (1) and (2) into Eq. (3): Cp ð1 AÞ þ ð

18 18 th tew Þ HR 0 fz minðfh1 ; fh2 Þ

ð3Þ

By analyzing the relevant parameters in (3), the passenger’s selection conditions for the passenger flow mode can be obtained.

4 Case Analysis 4.1

Background

Assume that NE is the passenger flow area subnet of the fast passenger network. According to the deﬁnition of the regional subnet in this paper, the maximum distance between any passenger flow nodes in the NE is 6 h journey (travel time, without medium conversion multiplication time). At present, the highest operating speed of China’s high-speed rail has three series of 300 km/h, 250 km/h and 200 km/h. The average travel speed of trains is greatly affected by the stop plan. The average travel speed is 217.3 km/h through the whole travel speeds of Beijing-Shanghai, Zhengxi, Hukun, Guiguang, Beijing-Tianjin, Xicheng and Chengmian. The above statistics are relatively rough and do not take into account the difference in the number of trains in each line. In order to make the study without loss of generality, the average travel speed of the fast passenger transport network is 200 km/h, from which it is determined that the average diameter of the passenger flow area subnet NE is 1200 km. The average fare of China’s fast passenger transport network is 0.4–0.6 yuan/km, this paper takes 0.45 yuan/km. Each passenger flow node in the NE belongs to the same fast passenger transportation network. It is assumed that the two passenger flow transmission modes are the same through the same route, and the travel mileage, the travel time in transit, and the total fare are basically the same. The average time value of passengers can be estimated based on China’s gross domestic product and the total number of employed

Selection Conditions Analysis of Passenger Transport Mode

275

people. In 2016, China’s GDP was 744.27 billion yuan, and the total number of employed people in the country was 776.03 million. According to the work of 22 days per month, 8 h per day, the time value of each hour can be estimated, and the rapid passenger transport network should be considered. The proportion of business travelers is higher, and the average time value of this paper is HR = 45 yuan/h. 4.2

Selection Condition Analysis

Although the following modes are the same for the direct and medium conversion modes, the intermediate conversion multiplication will inevitably need to stay at the transfer station, which objectively increases the expenditure of the generalized travel expenses. Satisfaction (3) is the basis for passengers to choose to transfer, so they can be classiﬁed to discuss the joint ticket discount rate (A), journey (L1 þ L2 ), transfer cost (th þ tew ), transit mode departure frequency (fh ), direct mode departure. Direct mode departure frequency (fz ) and other factors. Assume that the discount rate A of the joint ticket is 0, and the transfer cost is 1 h, 1.5 h, and 2 h, respectively. The corresponding driving frequency of the direct train and the transit train is as shown in Fig. 1.

Fig. 1. The relationship between transfer cost and train travel frequency without joint ticket discount rate

As can be seen from Fig. 1, when the discount rate (A) of the joint ticket is taken as 0, the passenger’s willingness to select the transfer mode can be increased by increasing the frequency of the transfer train. The higher the transfer time cost (th þ tew ), the higher the minimum travel frequency required to transfer the train corresponds to. When the transfer cost is large, the transfer frequency of the following cars in the transfer mode is much faster than the direct mode, and even exceeds the number of possible open rows. In Fig. 1, when the transfer cost is 2 h, the direct mode departure frequency is 8 times, and the matching transfer mode frequency reaches 72 times. It can be seen that when the discount rate (A) of the joint ticket is taken as 0,

276

Q. Zhang et al.

simply relying on increasing the frequency of the transfer to increase the passenger’s willingness to transfer, the scope of implementation is limited. Further discuss the relationship between the discount rate of the joint ticket (A) and the transfer time cost (th þ tew ) acceptable to the passenger. As mentioned above, the train operation mode based on the large station transfer is beneﬁcial to the effective utilization of the railway section passing capacity and the operation of the railway department. The railway department can establish a reasonable discount rate for joint tickets according to different journeys to enhance passengers’ willingness to transfer. From Fig. 2, the relationship between the different journey lengths (L1 þ L2 ), the joint ticket discount rate (A) and the passenger’s acceptable transfer cost (time, energy). Under the condition that the discount rate of the joint ticket is ﬁxed (0.6–1), the longer the journey, the higher the transfer time cost (th þ tew ) acceptable to the passenger.

Fig. 2. Relationship between transfer cost and joint ticket discount rate under different journey lengths

Discuss the relationship between the journey and the frequency of train travel under the condition that the joint ticket discount rate (A) is determined. According to the experience of foreign railway operation, the discount rate of joint ticket is generally 0.8–0.9. Discussion When A = 0.9, passengers are willing to choose the transfer mode of the direct and transit trains in the transfer mode under different travel conditions, as shown in Fig. 3. As can be seen from Fig. 3, when the journey is short, the discounted advantage of the joint ticket is lower than the transfer fee. At this time, the passenger’s willingness to travel is greatly affected by the frequency of travel in different modes. At the same time, the joint ticket discount can effectively offset the passenger’s transfer time cost. The longer the journey, the more obvious the advantages of the link discount. In contrast, the effect of increasing the frequency of the following vehicles in the medium-transition multiplication mode is not obvious. The numerical simulation results are shown in

Selection Conditions Analysis of Passenger Transport Mode

277

Fig. 3. The relationship between the frequency of direct and transit trains under different journey lengths (A = 0.9)

Figs. 4, 5, and 6. From the numerical simulation results, it can be seen that the railway department can flexibly determine the frequency of direct and transit travel according to the length of the journey and the discount rate of the joint ticket, so as to guide the passenger to choose the transfer.

Fig. 4. Comparison of train driving frequency in direct and transfer mode

Fig. 5. Direct and transfer mode train travel frequency comparison

Through the above analysis of the numerical relationship between the components of the general travel expenses such as the discount rate of the joint ticket, the frequency of departure, and the cost of the transfer, the railway department can adjust the value of the three to guide the passengers to choose travel and design the passenger products to get the goal.

278

Q. Zhang et al.

Fig. 6. Direct and transfer mode train travel frequency comparison

5 Summary This paper studies the passenger flow mode of the fast passenger transport network. It is considered that except for some large passenger flow nodes with obvious political and economic status, the passenger flow mode should be selected based on the passenger flow area subnet. Combined with relevant research results, the deﬁnition of passenger flow area subnet is given, and the characteristics and selection principles of transfer nodes in regional subnet are discussed. The composition of the general travel expenses of the direct and transit trains is discussed, and the numerical relationship between the components is analyzed. Based on the above analysis, the conditions for passengers to choose the middle conversion mode are analyzed. Combining speciﬁc regional subnets and further considering the restrictions on transportation resources such as section and station capacity, it is the focus of the next step to rationally formulate the ratio of direct and transit modes. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015,2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Li, Y., Li, H., Wang, Y., Yu, X.: Study on transportation organization of Beijing-Shanghai high-speed railway under transfer mode. Railway Transp. Econ. 39(01), 46–50 (2017) 2. Zhang, H.: Discussion on train operation modes of high-speed railways under railway network. Railway Transp. Econ. 37(09), 25–28 (2015) 3. Xiang, J.: Research on the passenger transportation mode of high-speed railway network. Sci. Technol. Enterp. (12), 26 (2015) 4. Yan, Y., Cui, Y.: Study on transportation mode of passenger flow in high speed railway abroad. Railway Transp. Econ. 34(08), 29–33 (2012)

Selection Conditions Analysis of Passenger Transport Mode

279

5. Wang, S., Zhao, P.: Analysis of conditions of schemes for choice between direct mode and transfer mode of express traveler network. Syst. Eng. 29(03), 47–52 (2011) 6. Qi, Zhang: Train operation mode decision method of dedicated railway passenger Line. J. DaLian Jiaotong Univ. 31(06), 6–10 (2010) 7. Jing, X.: Study on reasonable combination of conversion, multiplication and direct modes in passenger dedicated lines. Beijing Jiaotong University (2009) 8. Zuo, D.: Research on optimization of passenger train operation plan of railway rapid passenger transport network. Southwest Jiaotong University (2010) 9. Schöbel, A.: Line planning in public transportation: model and methods. OR Spectrum 34(3), 491–510 (2012)

Study on Optimal Allocation of Rail Transit Capacity Based on Utility of Passenger Flow Transfer and Loss Qiangfeng Zhang1,2,3,4, Shaoquan Ni1,2,3(&), Gaoyong Huang4, and Wentao Li1 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 School of Information Science and Technology, Southwest Jiaotong University, Chengdu 610031, China

Abstract. There are imbalances in the scale of passenger flow in different periods of the rail transit operation day, which can be generally divided into peak, flat peak and low peak periods. Due to the influence of maximum transport capacity and other factors, the passenger flow in each period of rail transit will have a loss effect, and passenger flow transfer effect will occur in different time periods, resulting in a random dynamic evolution of the passenger flow scale distribution during the period. In view of this, based on the selection characteristics of passenger flow random utility maximization, this paper proposes a passenger flow transfer quantity determination method based on passenger flow transfer utility, establishes passenger flow transfer model based on random utility theory and passenger flow transfer probability function, determines the transfer passenger flow in each time period, and then determines actual passenger flow at each time. On this basis, the evaluation system of capacity allocation schemes in each period is established, and the simulations are carried out through examples to analyze the advantages and disadvantages of the capacity allocation schemes in each period to provide decision-making basis for decision makers. Keywords: Passenger flow transfer Capacity conﬁguration

Utility function Transition probability

This research has been supported by the National Key Research and Development Program of China (2017YFB1200702, 2016YFC0802208), National Natural Science Foundation of China (Project No. 1703351), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), and Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007), Fundamental Research Funds for the Central Universities (2682017ZDPY04, 2682017CX022, 2682017CX018), and the Science and technology research and development plan of Beijing-Shanghai high speed railway. © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 280–297, 2019. https://doi.org/10.1007/978-3-030-04582-1_33

Study on Optimal Allocation of Rail Transit Capacity

281

Due to the imbalance of passenger flow scale in different time periods, rail transit is often in peak hours, flat peak hours and low peak hours. Generally speaking, during peak hours, the transportation capacity is tight, and it is difﬁcult to meet the travel needs of all travelers during this period due to the ability and the like. At this time, passengers will dynamically adjust their travel behavior according to the changes in transport service capacity and level, resulting in passenger flow transfer [1] and loss [2] effects. Passenger flow in each period will also evolve randomly and dynamically with the influence of influencing factors. At present, relevant research mainly focuses on qualitative or unqualiﬁed passenger flow forecasting and distribution. There are few related studies on passenger flow dynamic adjustment such as passenger flow transmission and loss due to transportation service capacity and level, resulting the distortion in existing passenger flow forecasting, etc. On the basis of the existing passenger flow forecast, the study of passenger flow and loss can accurately determine the actual scale of passenger flow in each period, improve the passenger flow forecasting and distribution system, and provide decision-making basis for the optimal allocation of transportation enterprises. It is beneﬁcial to improve the quality of transportation services and the economic beneﬁts of transportation enterprises and have high academic research and practical application value.

1 Deﬁnitions and Assumptions Affected by many factors, the distribution number of rail transit passenger flow periods will evolve randomly and dynamically with the change of influencing factors, and passengers will dynamically adjust their travel behavior according to the changes of transport service levels. Because of the difference in passenger perception of cost, the selection feature that maximizes the utility of randomness is typically represented by the multivariate Logit model [3–5]. Because the probability expression of the multivariate Logit model is dominant, and the model is fast and simple, it is widely used. This section is based on the maximum random utility chosen by passengers [6, 7]. The Logit model is used to analyze and model the passenger flow transfer. The following deﬁnitions and assumptions are made for the convenience of description: 1.1

Deﬁnition

(1) Source passenger flow For a certain period of time, the passenger flow that is expected to travel during that time period is called the source passenger flow of the time period, and can theoretically be regarded as the original predicted passenger flow. (2) Transfer of passenger flow Because of the transportation capacity problem, passengers can not travel in the expected time, the passenger will choose to travel in time that is relatively loose. This part of the passenger flow is called the transfer passenger flow, that is, the passenger

282

Q. Zhang et al.

who can’t ride in the desired time period and chooses other time slots, the process is called Transfer for passenger flow. (3) Loss of passenger flow After the passenger flow transfer in a certain period of time, the remaining passenger flow still exceeds the passenger flow capacity, and the excess passenger flow will be forced to abandon the travel demand. This part of the passenger flow is called the lost passenger flow. (4) Time to be transferred The duration of the expected travel time and the travel realization period of the transfer passenger flow is called the waiting time, and the special waiting time for the passenger flow that can be traveled in the desired time period is zero. 1.2

Assumption

The passenger flow transfer discussed in this paper is based on the problem that the passenger flow demand in a certain period is greater than the transport capacity, that is, the passenger can not travel in the expected time period due to the problem of transport capacity, and choose the travel time in the subsequent period. The problem of passenger travel time changes caused by passengers themselves and other reasons is not within the scope of this paper, and no speciﬁc analysis is made. For the convenience of description and research, the following assumptions are made regarding passenger flow transfer: (1) Passenger flow transfer does not cross the operating day hypothesis The transfer period of the transfer passenger flow can only be the time of the same operation day, and cannot be transferred to the time of other operation days. (2) Passenger flow transfer order hypothesis The transfer of the passenger flow is sequentially performed from the ﬁrst time period of the operation day, and the calculation is sequentially performed in the order of the time slots. (3) One-way assumption of passenger flow transfer The transfer of passenger flow is unidirectional, that is, the passenger flow can only be transferred to the subsequent time period of the expected travel time, and cannot be transferred in the previous time period. (4) Passenger flow transfer does not occur hypothesis When the source passenger flow in a certain period of time and the total amount of passenger flow transferred to the time period are not greater than the transport capacity that can be provided during the time period, the passenger flow does not shift during the time period.

Study on Optimal Allocation of Rail Transit Capacity

283

(5) Unconstrained hypothesis of passenger flow transfer period When the passenger flow is transferred, it can be transferred to the subsequent time periods of the expected travel time until the last time of the operation day, except that the transfer probability changes. (6) Passenger flow transfer termination assumption For any transferred passenger flow, when it is transferred to a certain period of time, its transfer is terminated. Even if there is still a passenger flow transfer phenomenon during the transfer period, this part of the passenger flow is no longer transferred. (7) Deterministic assumptions for transferring passenger flows When the passenger flow in a certain period of time is transferred to the subsequent time period, the passenger flow transferred (transferred to other period) during the time period is the source passenger flow of the current time period.

2 Analysis of Associated Utility Functions In order to make different types of utility functions comparable, in the utility function design, whether it is associated time utility or associated trafﬁc utility, the maximum value is 1 (normalized processing). 2.1

Associated Time Passenger Flow Transfer Utility Function

For the transfer of passenger flow, the waiting time is an important factor affecting the passenger transfer time. In general, passengers expect to travel as quickly as possible, and the waiting time is as short as possible. Therefore, for the transfer of passenger flow, the longer the waiting time, the lower the utility; the shorter the waiting time, the higher the utility. The passenger flow transfer utility function that is transferred from the i period to the j period can be expressed as: U ki;j ¼ eaki;j ; i ¼ 1; 2; ; N

0 ki;j \Ni

ð1Þ

In the function: i: any time of the operation day; ki;j : the number of transfer periods in which the passenger flow is transferred from the i period to the j period, and ki;j ¼ j i, ki;j ¼ 0 indicates that the passenger stays in the current period i; N: the number of time slots divided by the operation day; a: Utility function adjustment factor

284

Q. Zhang et al.

Fig. 1. Curve of utility flow of associated time passenger flow transfer

As can be seen from Fig. 1: (1) When the time to be transferred or the time of transfer is 0, the transfer effect of the passenger flow is the largest, or the passenger flow is transferred to the source time period, that is, the most effective during the expected travel time; (2) The transfer effect of the transfer passenger flow gradually decreases with the increase of the waiting time (the number of transfer periods); (3) The adjustment coefﬁcient a can be used to adjust the attenuation sensitivity of the passenger flow transfer utility associated with the time, the larger a is, the greater the attenuation. 2.2

Association Ability Passenger Flow Transfer Utility Function

The utility of passenger flow transfer is closely related to the remaining capacity of the time-shifting capacity. The greater the remaining capacity of the time-shifting capacity, the higher the utility of the passenger flow into the utility; the lower the remaining capacity, the smaller the utility of the passenger flow. Therefore, the utility of the passenger flow transfer during this period can be measured by the remaining capacity of the period. The passenger flow transfer utility associated with passenger trafﬁc can be divided into two situations. One is the utility of the passenger flow into a certain period of time, and the other is the utility of the transfer of a certain event. The analysis method is similar. This section analyzes the utility of passenger flow into a certain time period. Assuming two periods i and j (i\j), before the passenger flow in the i period is transferred to the j period, the passenger flow in the j period includes the source passenger flow (original predicted passenger flow) of the time period and the passenger flow transferred to the i period in the previous period of the j period. (j period accepts the passenger flow of the period before the i period). In this case, the passenger flow of 0 j is represented by Ci;j , then:

Study on Optimal Allocation of Rail Transit Capacity

0

Ci;j ¼ Cj þ

i1 X

Cm;j ; 1\i j N

285

ð2Þ

m¼1

Among them, Cm;j represents the passenger flow from the m period to the j period, and Cj is the predicted trafﬁc in the j period. Then, the passenger flow transfer effect 0 from the i time period to the j time period, that is, the utility UðCi;j Þ of the passenger flow in the j period accepting the i time period is: 0 0 B e U Ci;j ¼ B @

½bð

1þe

0 i;j j 1Þ Cmax C

0 C ½bð ji;j 1Þ Cmax

1ki;j C C A

ijN

ð3Þ

In the formula: j Cmax : Maximum passenger flow capacity of time period j (maximum passenger capacity), (unit: person); b: Correlation ability utility adjustment factor; ki;j : Association time (number of time slots) penalty factor, ki;j ¼ j i þ 1; 0 Ci;j : The passenger flow of the j period before the transfer of the passenger flow in the i period to the j period (unit: person).

The effect of the related ability passenger flow transfer utility change, the influence of the adjustment factor on the transfer effect of the associated capacity passenger flow and the impact of the penalty factor on the transfer effect of the associated capacity passenger flow are analyzed, as shown in Figs. 2, 3 and 4.

Fig. 2. Correlation ability passenger flow transfer utility curve

286

Q. Zhang et al.

Fig. 3. Effect of adjustment coefﬁcient on the utility of related ability passenger flow transfer

Fig. 4. Effect of penalty factor on the utility of associated capacity passenger flow transfer

It can be seen from Figs. 2, 3 and 4: (1) The transfer effect of passenger flow transferred to a certain period of time is positively correlated with the passenger flow capacity during this period. The greater the passenger flow capacity, the greater the utility of passenger flow transfer, that is, the stronger the ability to accept the transfer passenger flow during this period; the smaller the passenger flow capacity, The smaller the utility of the passenger flow transfer, the weaker the ability to accept the transfer of passenger flow during that time.

Study on Optimal Allocation of Rail Transit Capacity

287

(2) The transfer effect of passenger flow transferred to a certain period of time is negatively correlated with the demand for passenger flow during this period. The larger the demand for passenger flow, the smaller the utility of passenger flow transfer, that is, the weaker the ability to accept the transfer of passenger flow during this period; the smaller the demand for passenger flow The greater the utility of passenger flow transfer, that is, the stronger the ability to accept the transfer of passenger flow during this period. (3) For the same period of passenger flow demand and transport capacity, regardless of whether the transport capacity is the same, the passenger flow transfer utility is the same, that is, the ability to accept the transfer passenger flow is the same. (4) The correlation ability utility adjustment factor has a signiﬁcant impact on the passenger flow transfer effect. The larger the adjustment factor b, the steeper the curve corresponding to the maximum transport capacity in a certain period of time, and the sensitivity of adjusting the passenger flow demand to the utility effect. When the actual passenger flow demand approaches or exceeds the transport capacity, the passenger flow transfer effect is rapidly attenuated, that is, the ability to accept the transfer passenger flow is rapidly reduced during the period; conversely, when the actual passenger flow and the transport capacity differ greatly, the passenger flow transfer utility increases. The ability to accept passenger flow transfer during the time period is stronger. (5) Correlation time (number of time slots) The passenger flow transfer penalty factor has a certain influence on the passenger flow transfer utility, keeping other conditions unchanged, and the longer the waiting time is (the more the number of transfer periods), the smaller the passenger flow transfer utility is, that is, the acceptance ability of the passenger flow during this period is weaker.

3 Passenger Flow Transfer Model Based on Stochastic Utility Theory 3.1

Passenger Flow Transfer Utility Function

It is assumed that the utility of the originating passenger of the period i to shift to the j period is Ui;j (i j N), where Kij ¼ 0 indicates that the current period i is left, and Kij 6¼ 0 indicates that the passenger flow is transferred to the j period subsequent to the i period. In addition, Vij is the passenger flow transfer utility, ni;j is the random deviation, then the passenger flow utility function from the i period to the j period is: Ui;j ¼ Vi;j þ ni;j ; ni;j 2 ½0; 0:05

ð4Þ

0 Vi;j ¼ cU ki;j þ ð1 cÞU Ci;j i j N ki;j ¼ j i

ð5Þ

288

Q. Zhang et al.

In the formula: Uðki;j Þ: associated time (number of time slots) passenger flow transfer utility function; 0 UðCi;j Þ: associated ability passenger flow transfer utility function; c: weight factor. 3.2

Passenger Flow Transfer Probability Function

When the actual passenger flow demand during a certain period is greater than the passenger flow capacity during the period, the remaining passenger flow will be transferred to a subsequent period or abandoned. The transfer probability can be determined by the transfer probability, and the transfer probability of the passenger flow is directly affected by the transfer effect of the passenger flow. The probability of the passenger flow transition from the i period to the j is deﬁned as the ratio of the passenger flow transfer utility of the i period to the j period and the total utility of the i passenger flow transfer. Therefore, the probability of passenger flow transition from the i period to the subsequent j period is: Ui;j pi;j ¼ P ijN n Ui;m

ð6Þ

m¼i

In the formula: pi;j : transition probability of passenger flow from i period to j period. 3.3

Actual Passenger Flow During the Period

According to the formula (6), the transfer of the passenger flow in any one time period can be obtained in turn. It is known from the 1.2th assumption that the passenger flow transferred to a certain period of time no longer shifts. Therefore, the passenger flow transferred at this time is the source passenger flow of the time period. If the calculation period is the i period, the passenger flow of the i period to the j period is: 0

Ctransfer;i;j ¼ Ci pi;j i j N

ð7Þ

In the formula: 0

Ci : The sum of the originating trafﬁc before the passenger flow in the i period and the trafﬁc volume transferred to the i period before the i, the unit: person; Ctransfer;i;j : The trafﬁc of the i period to the j period. It is assumed that the actual passenger flow at any time i after the transfer of the passenger flow is obtained by three parts, that is, the sum of the remaining passenger flow after the transfer of the originating passenger flow (original predicted passenger flow) and the passenger flow transferred to the time period can be expressed as:

Study on Optimal Allocation of Rail Transit Capacity

Cstay;i ¼ Ctransfer;i;i þ

i1 X

Cm;i ; i ¼ 1; 2; . . .; N

1iN

289

ð8Þ

m¼1

If the passenger flow remaining in this period is greater than the maximum passenger flow capacity during the period, it is assumed that the source passenger flow exceeds the delivery capacity, and the passenger flow loss is: Cdrop;i ¼

Cstay;i Cmax;i 0

Cstay;i [ Cmax;i ; i ¼ 1; 2; . . .; N Cstay;i Cmax;i

ð9Þ

Therefore, the actual passenger flow for any time period i can be obtained, namely: Cleft;i ¼ Cstay;i Cdrop;i0 i ¼ 1; 2; . . .; N

ð10Þ

From the above analysis, the total amount of loss of passenger trafﬁc on the operating day can be reached, namely: Ctot;drop ¼

N X

Cdrop;i

ð11Þ

i¼1

In the formula: Cdrop;i : The number of passengers who abandon their travel at any time period i, that is, the loss of passenger trafﬁc, people; Cmax;i : The maximum passenger flow volume in any period i, person. 3.4

Passenger Flow Transfer Performance Evaluation Indicators

It can be seen from the foregoing analysis that under the premise of predicting the passenger flow and the maximum passenger flow volume in each period, the capacity allocation scheme will have a greater impact on passenger flow transfer and loss, which can directly affect the beneﬁts of transportation companies and passengers. In order to evaluate the pros and cons of the capacity allocation schemes in each period, this paper establishes the number of transition period per person of passenger flow and passenger flow loss rate indicators for evaluation. 1. Number of transition periods per person The number of per person transfer periods is deﬁned as the ratio of the total number of transfer hours of the transfer passenger flow to the total predicted passenger flow on the operation day. This indicator reflects the utility of passenger flow, that is, the quality of service. The fewer the number of transition periods per person, the higher the probability of trip passenger flow in the expected time period, the higher the utility, the higher the satisfaction of passenger flow, and the better the quality of transport service. The number of transition periods per person can be expressed as:

290

Q. Zhang et al. N1 P

nave ¼

N P

i¼1 j¼i þ 1

ki;j Ctransfer;i;j

N P m

ð12Þ Cm

In the formula: nave : The number of transition periods per person; Ctot : The sum of the predicted trafﬁc for each time period, person. 2. Deﬁnition of passenger flow loss rate After the passenger flow in a certain period of time, if the remaining passenger flow during the time period is still greater than the maximum transmission capacity of the time period, the passenger flow exceeding the transmission capacity portion will be unable to travel, and the transportation enterprise will lose the passenger flow. For transportation companies, they hope to maximize the beneﬁts, and hope to transport passengers as much as possible within the scope of their capabilities, and reduce passenger flow losses as much as possible. Deﬁne the passenger flow loss rate as the ratio of passenger flow (loss passenger flow) and total predicted passenger flow that are abandoned during the operation day, that is: N P

Cdrop;i Rdrop ¼ i¼1N P Cm

ð13Þ

m

According to the above two indicators, the advantages and disadvantages of the capacity allocation can be evaluated, which can be used to optimize the capacity allocation in each period.

4 Examples and Simulation Analysis 4.1

Simulation Scene Settings

According to the above analysis, the proposed scheme can evaluate the rationality of the capacity allocation in each period. On the one hand, the economic utility of the transportation enterprise and the satisfaction of passenger travel are measured by the number or percentage of passengers lost (abandoned travel); on the other hand, per capita The number of transfer periods is used to evaluate the quality of transportation services and passenger satisfaction. The passengers who need to transfer choose which time period is affected by the waiting time and the maximum passenger capacity of each time slot. Therefore, it is necessary to know the division of the time of day, the waiting time of the passenger, the

Study on Optimal Allocation of Rail Transit Capacity

291

maximum passenger capacity per time period, and the predicted passenger ﬂow set. In this section, the passenger travel time is divided into 16 time slots, and the interval is 6:00–22:00 (as shown in Table 1 below). The time-predicted passenger trafﬁc in the table takes the mean value of the maximum passenger volume in the corresponding period of the corresponding sample 2–10. For the convenience of comparative analysis, the maximum passenger capacity of the sample 1 is assumed to be equal to the predicted trafﬁc volume, and the simulation is shown in Fig. 5. The “sample 1” in 12 is the comparison of the simulation results of the simulation results under the condition of the maximum passenger capacity in the sampling period. The value of the sample data is based on the bimodal model of the distribution characteristics of the passenger line of the passenger line. In this section, the simulation setting model parameter a is 3, b is 30 and c is 0.1. Table 1. Passenger flow demand and sample of maximum passenger flow capacity during the period 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

6:00– 7:00

7:00– 8:00

8:00– 9:00

9:00– 10:00

10:00– 11:00

11:00– 12:00

12:00– 13:00

13:00– 14:00

14:00– 15:00

15:00– 16:00

16:00– 17:00

17:00– 18:00

18:00– 19:00

19:00– 20:00

20:00– 21:00

21:00– 22:00

1048

1875

2187

3138

2587

2538

2212

1870

2425

2542

2615

3211

1767

1468

1157

924

1129

1768

2167

3217

2534

2344

2145

1850

2354

2534

2898

3535

1456

1298

1023

985

1054

2107

1823

2561

2456

2457

2454

1864

2633

2589

2342

3254

1764

1556

1125

950

952

1865

1956

3214

2575

2689

1957

1836

2653

2567

2754

3452

1564

1320

1186

967

1068

1973

2176

3334

2753

2667

1873

1884

2544

2545

2757

3214

1754

1629

1239

967

968

1567

2341

2715

2465

2534

2314

1895

1966

2469

2455

2986

1542

1246

1089

870

1079

2353

2527

3344

2654

2465

2124

1905

2455

2566

2764

2797

1674

1528

1258

968

1167

1642

1965

3451

2642

2436

2354

1856

2356

2415

2464

3215

2143

1523

1023

824

983

2165

2376

3167

2543

2575

2341

1843

2316

2645

2556

3315

2143

1687

1259

957

1034

1437

2354

3241

2657

2677

2345

1894

2546

2545

2546

3132

1865

1429

1207

824

1048

1875

2187

3138

2587

2538

2212

1870

2425

2542

2615

3211

1767

1468

1157

924

4.2

Simulation Results and Analysis

As can be seen from Fig. 5, the maximum passenger capacity of each period of sample 1 is equal to the predicted passenger flow rate; the maximum passenger flow (maximum passenger capacity) of sample 2 to sample 10 is randomly taken around the mean value of the time period, which basically conforms to the bimodal type. The trend of change. For the convenience of comparative analysis, it is assumed that the mean passenger flow (maximum passenger capacity) of the time period of sample 2 - sample 10 is taken as the maximum passenger flow volume (maximum passenger capacity) of each time period of sample 1, and the source passenger flow rate of each time period is assumed. That is, the original predicted passenger flow is equal to the maximum passenger flow (maximum passenger capacity) of each period of sample 1. Figure 6 shows the comparison of the maximum passenger flow volume of each sample throughout the day with the original predicted passenger flow. Of the samples given, sample 6 had the smallest total passenger flow throughout the day. The maximum passenger flow volume of sample 1 is equal to the predicted passenger flow due

292

Q. Zhang et al.

Fig. 5. Predicts passenger flow and sample maximum passenger flow

Fig. 6. Comparison of total passenger flow and total passenger flow during the whole day

to the set time period. Therefore, the total passenger flow volume of the whole day is equal to the total passenger flow forecast for the whole day. Figure 7 shows the passenger flow transferred from the previous period in each period. It can be seen that for sample 1, since the maximum passenger flow in each time period is equal to the predicted passenger flow in the time period, the maximum passenger flow volume in each time period can satisfy the demand of the predicted

Study on Optimal Allocation of Rail Transit Capacity

293

Fig. 7. The amount of passenger flow transfer (acceptance) in each period

passenger outflow line in the time period, and no passenger flow transfer occurs. Since this paper assumes that the passenger flow will be transferred to each time period, the ﬁrst time of the sample 2–10 has no passenger flow transferred from the previous time period, and the other passenger trafﬁc received in the other time periods is related to the predicted passenger flow during the time period and the maximum passenger flow volume during the time. Figure 8 is a time-lapse predicted passenger flow loss rate. It can be seen from the ﬁgure that the sample 1 has the maximum passenger flow capacity in each period to meet the predicted passenger outflow demand, so there is no passenger flow loss. The passenger flow loss rate is relatively large during the period 13-period 16 because the cumulative passenger flow from the previous period to the subsequent period is more. According to the assumption of this paper, since the passenger flow in the previous period is accepted in the subsequent period, in the subsequent period, the source of the forecasted passenger flow is more likely to abandon the ride, resulting in a large loss of passenger flow. Figure 9 shows the proportion of passenger trafﬁc that is transferred to the subsequent period in each period of time to the predicted passenger flow during that period. It can be seen from the ﬁgure that for the pre-continuation period, since the unidirectionality of the passenger flow transfer is assumed in this paper, the probability that the source passenger flow will be transferred to the subsequent time period in the previous period is greater; and for the subsequent time period, the maximum passenger capacity sample under different time periods The space transferred to each subsequent period is relatively small, but it is known from the foregoing analysis that the passenger flow loss rate is relatively high in subsequent periods; since the originating passenger flow in the last period cannot be transferred to the subsequent period, therefore, the transferred passenger flow is 0.

294

Q. Zhang et al.

Fig. 8. Time rate of passenger flow loss

Fig. 9. Percentage of passenger flow in the period from the time period to the subsequent period

Figure 10 shows the proportion of total passenger flow transferred throughout the day for each sample. It can be seen from the ﬁgure that sample 1 can meet the predicted passenger flow travel demand due to the maximum passenger flow volume in each period. Therefore, no transfer occurs. Samples 5 and 7 have the largest passenger flow due to the maximum passenger flow during the whole day. The difference in transport

Study on Optimal Allocation of Rail Transit Capacity

295

Fig. 10. Total proportion of passenger flow transferred throughout the day

volume results in different transfer passenger flows; although the total passenger flow of sample 6 is much lower than that of sample 5 and sample 7 throughout the day, the total amount of passenger flow transfer is only slightly higher than sample 5 and sample 7, this shows the setting scheme of the maximum passenger flow volume during the period, that is, the capacity conﬁguration scheme directly affects the passenger flow transfer. Figure 11 is a comparison of the number of per person transfer periods for each sample. It can be seen that the sample 1 has the maximum passenger flow capacity to meet the passenger outflow demand, so there is no passenger flow transfer; although the sample 9 has the largest total passenger flow, the per person transfer time is the largest, the transportation service quality is the lowest, and the passenger satisfaction is the lowest. Except for sample 1 without transfer, the sample 7 has the lowest number of transfer periods per person; compared with Fig. 12, it can be found that although sample 6 has a small number of transfer periods per person, the loss rate of passenger flow is large, and the economic beneﬁts of transportation enterprises are poor. Figure 12 shows the loss rate of passenger flow throughout the day. It can be found that the maximum passenger flow rate of the sample 6 is much smaller than the full-day forecast passenger outflow demand, and the passenger flow loss rate is the largest; as can be seen from Fig. 6, although the maximum passenger flow of the whole day of samples 5 and 7 is relatively close. However, the difference in passenger flow loss rate is relatively large throughout the day. The maximum passenger flow rate of sample 5 is slightly smaller than that of sample 7, but the loss rate of passenger flow throughout the day is lower than that of sample 7, which is due to the difference in maximum passenger flow during the two samples. From the simulation data in this section, when the maximum passenger flow volume in the existence period cannot meet the passenger outflow demand, the passenger flow will be transferred and lost. Although the total passenger flow volume during the

296

Q. Zhang et al.

Fig. 11. Number of per person transfer periods

Fig. 12. Loss rate of passenger flow throughout the day

whole day can meet the demand for all-day passenger outflows, it does not guarantee the lowest passenger flow loss rate or the minimum passenger transfer time. It is necessary to optimize the service flow during the conﬁguration period, that is, the passenger flow capacity to achieve a win-win situation between service quality and the economic beneﬁts of transportation enterprises.

Study on Optimal Allocation of Rail Transit Capacity

297

5 Conclusion This paper takes the transportation capacity and forecast passenger flow in each period as the object of investigation, analyzes the correlation time utility function and the correlation ability utility function, and obtains the relationship between the size of the passenger flow transfer utility and the influencing factors. The passenger flow transfer model based on the random utility theory is established. And the passenger flow transfer probability function, determine the flow and loss of passenger flow in each time period, and then determine the actual passenger flow in each time period; establish the evaluation system of the time capacity allocation plan with the number of per capita transfer time and the loss rate of passenger flow; simulate by example The evaluation and analysis of the transportation capacity allocation scheme (passenger trafﬁc volume) in different periods provides decision-making basis for decision-makers. Decisionmakers can reasonably determine the capacity allocation schemes for each period according to actual needs, thereby improving service quality and economic beneﬁts. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Luo, F.: Uncertainty analysis of urban rail transit passenger flow forecast. Changan University (2016) 2. Bao, J.: Discussion on the mode of medium and long distance trains based on economic beneﬁt calculation. Integr. Transp. 40(03), 61–65 (2018) 3. Guo, J.: Urban rail transit network passenger flow control method. Beijing Jiaotong University (2016) 4. Zhang, Y., Yao, E., Liu, S., Cai, C.: Modeling and application of half compensation path selection for urban rail transit passengers. J. Railway 40(02), 1–7 (2018) 5. Yi, H., Chen, J.: Research on optimization of intercity railway train operation plan based on logit price response function. J. Transp. Eng. Inf. 16(02), 28–35 (2018) 6. Li, Y.: Research on passengers’ ticket purchase behavior under the condition of high speed railway parallel trains. Beijing Jiaotong University (2017) 7. Cai, C., Yao, E., Zhang, Y., Liu, S.: Forecast of passenger flow distribution between urban rail stations based on AFC data. China Railway Sci. 36(01), 126–132 (2015) 8. Xu, P.: Travel demand and passenger time-space distribution of passenger dedicated lines. Beijing Jiaotong University (2012)

Connotation and Direction of Urban Rail Transport Integration Tingting Wu1, Shuming Huang2, Shaoquan Ni3(&), and Rong Kuang1 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu, China [email protected], [email protected] China Railway SIYUAN Survey and Design Group Co., Ltd., Wuhan, China [email protected] 3 National Railway Train Diagram Research and Training Center, Chengdu, China [email protected]

Abstract. Aiming at the planning asynchrony of urban rail transport and cities, the insufﬁcient transferring capacity of urban rail system during rush hours, and the low connection between rail and ground public trafﬁc, this paper comes up with planning integration of urban and rail transport, connection integration of urban rail transport and ground public trafﬁc, and organization integration of operation plan, probing into the points and difﬁculties of planning, construction and operation period, and forecasting the trend of urban rail transport integration. Keywords: Urban rail transport Planning integration Connection integration Organization integration

1 Introduction By the end of 2016, China has built and put into operation more than 4,000 km of rail transport lines and approved rail transport construction plans over 40 cities, with a total mileage exceeding 8,600 km [1]. With the vigorous development of urban rail transport, problems have gradually emerged in the operational urban rail transport lines, such as the low match between the size of lines and the setting of stations; congestion in the station and long waiting time for transfer, and the poor connectivity between rail and other modes of transport. The key lies in the lack of consideration for the integration of urban rail transport. Urban rail transport integration refers to the formation of an urban transport mode with rational division and organization, full sharing of resources and orderly and close links. Integration is embodied in all stages of urban rail transport development, including the planning integration of urban and rail transport during planning stage, the connection integration of urban rail transport and ground public trafﬁc during design and construction stages, and the organization integration of operation plan during operation stage.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 298–304, 2019. https://doi.org/10.1007/978-3-030-04582-1_34

Connotation and Direction of Urban Rail Transport Integration

299

2 Planning Integration of Urban and Rail Transport 2.1

The Relationship Between Urban Rail Transport and Urban Planning

Urban rail transport planning usually appears as an integral part of urban transport planning, which influences the overall urban planning and other sub-planning by driving the flow of population. In short, urban planning guides urban rail transport planning, while urban rail transport planning serves and affects urban planning (Fig. 1).

Fig. 1. The map of relation of urban rail transport and urban planning

With the intensiﬁcation of urbanization, the population density is increasing, and the land use will also become more intense. The city’s demand for public transport, especially rail transport, is getting increasingly higher. However, the present rail transport in many cities is not closely integrated with urban planning, but presents a separated and unrelated state. The relationship between them should not only stay at the Leader-Member Level, but also need to establish a mutual feedback mechanism between the two, and realize the integration of urban rail transport and urban planning, so as to get the higher rate of land use and more compact layout, as well as effectively reduce the pressure of public transport and reduce the cost of urban transport. 2.2

The Points of Planning Integration

Urban rail transport planning mainly relies on urban master planning. It serves urban planning and guides urban planning. The planning integration of urban transport and urban mainly includes the following aspects. (1) The integration of urban rail transport layout and urban development axis The role of urban rail transport in urban development is mainly reflected in that urban rail transport can attract and gather public transport passenger flow, promote the development and utilization of land resources along the line, and enhance the industrial and commercial value along the line. In order to realize the integration of rail transport layout and urban development axis, it is necessary to carry out the layout of urban rail transport along the main axis of urban development or the main passenger flow channel, considering the main development direction and

300

T. Wu et al.

passenger flow direction of the city, so as to connect the passenger flow distribution points and important stations along the line in series and shorten their time distance. (2) The integration of rail transport capacity and urban development scale Before the reform and opening up, the urbanization level of China was low and its development was slow. After the reform and opening up, the urbanization process has been in a rapid development stage. Since the time and depth of rail transport planning is different from that of urban master planning, which means that the existing urban rail transport capacity would might not be able to satisfy the actual transport demand of passenger flow and the scale of stations. With the continuous improvement of urban planning, the continuous dispersion and aggregation of travelling population and the changing peak period might also appear, which will affect the supply and demand relationship of urban rail transport. The pressure of urban rail transport will change flexibly with the development of urban planning, but during the initial planning period, various possibilities of urban development scale must be taken into account to ensure that urban rail transport has certain adaptability to urban development scale. (3) The integration of urban rail transport and urban complex Gigantism and pluralism will be the development trend of urban architecture. In order to realize the integration of urban rail transport and urban complex, comprehensive consideration should be given to the urban and architectural attributes of the complex, the spatial attributes of trafﬁc links, and the intensity of passenger flow. Especially, the role of underground space for rail transport stations should be fully taken into account, so as to provide urban complexes with convenient and efﬁcient transport services under the insurance of safety and noninterference. Therefore, it can not only attract more potential passenger flow by virtue of urban complex, reduce the pressure of urban ground trafﬁc development, but also bring business opportunities for commerce, catering and entertainment along the urban rail transport, improving the economic beneﬁts of urban complex and selfeconomic value. 2.3

The Development of Planning Integration

The future development direction of the planning integration of urban and rail transport lies in the in-depth study of the urban master plan, and comprehensive consideration of the objectives and requirements of the upper-level planning such as trafﬁc planning and socio-economic development planning for rail transport. Considering the city’s ﬁnancial capacity, the level of rail transport construction and management capacity, it should combine with the construction focus and direction of a city in a certain period of time, and choose appropriate construction scale and projects within the scope of urban ﬁnancial resources, so as to give full play to the role of urban rail transport and promote the realization of urban development goals.

Connotation and Direction of Urban Rail Transport Integration

301

3 Connection Integration of Urban Rail Transport and Ground Public Trafﬁc 3.1

Necessity of Connection Integration

Urban rail transport is the development trend of modern mass transport, and gradually occupies a dominant position in the public transport of major cities. Since the service scope and accessibility of urban rail transport is small, it is necessary to cooperate closely with the ground passenger transport mode to complete the urban passenger transport task. Especially, since urban rail transport and ground bus system (including conventional bus and BRT) have the vast majority of urban public transport passenger volume, the organic connection between them plays a decisive role in ensuring the normal operation of the entire urban public transport network and the efﬁciency of urban passenger flow. 3.2

The Points of Connection Integration

The realization of the connection integration between urban rail transport and ground bus network calls for the design of urban rail transport and ground bus network integration, the transfer facilities integration and operation management integration, so as to brings the “trinity” among network, facilities and operation of urban rail transport and ground bus system. The above three integration points are analyzed as follows. (1) The integration of trafﬁc network In order to avoid ineffective competition, rail transport lines should be used as the backbone at the line level to reduce the repetition of parallel ground buses and increase the number of intersecting lines to connect the transferring passenger flow, at the site level, though, some bus routes within the direct attraction area of station should be canceled or adjusted, while some within the indirect attraction area would be added, so that the passenger flow can be smoothly and reasonably carried out among the two parties to ensure the comprehensive beneﬁts of the public transport network. (2) The integration of transfer facilities The integration of transfer facilities refers to the reasonable arrangement and connection between urban rail transport and ground bus facilities, which makes the transfer connection smooth and efﬁcient. Usually, the ground bus lines and rail transport stations are gathered as far as possible, and the facilities are arranged as compact as possible, so as to reduce the passenger transfer journey and time. In addition, the start and end points of urban rail transport are often located in the border area between the city center and suburbs, which has a wide range of radiation attraction, and is conductive to the uniﬁed construction of large-scale transfer hub. (3) The integration of operation and management The integration of operation and management is to establish a uniﬁed and coordinated operation and management mode between urban rail transport and ground public transport, which matches operation dispatch, scheduled shifts, connection supply and demand, and fare systems to ensure the matching of the two in dispatch and operation organization.

302

3.3

T. Wu et al.

The Development of Connection Integration

With the gradual enlargement of rail network and the gradual increase of station in major cities in China, the integration of urban rail transport and ground transport will enter the stage of which the rail transport line to be the main body, while the ground bus line to be the supplement, the mode of which the ground bus mainly delivers passengers for rail transport. Therefore, the development trend of connection integration is that there are fewer bus lines in the central urban area where the rail transport network is densely distributed, while the bus lines in transfer hubs and peripheral areas are densely distributed, which attract passengers in the form of circular lines around rail transport stations.

4 Organization Integration of Urban Rail Operation Plan 4.1

Core Contents and Relationship of Operation Plan

Urban rail transport operation plan is a technical document which is used to direct urban rail trains to realize passenger displacement service safely and conveniently. Under the macro-strategy requirements and ﬁxed facilities and equipment conditions of urban rail transport operation, a reasonable operation plan can not only meet the travel needs of urban passengers, but also give full play to the capacity of facilities and equipment along the line and station, so as to save operating costs and improve the service level and operational efﬁciency of urban rail transport. The core contents of urban rail transport operation plan include train operation plan, train diagram and vehicle operation plan. These plans determine the running state of train at various times, stations and sections. The train operation plan is a frame plan that transforms passenger flow into train flow. On this basis, according to the relevant rules of transportation safety, the train diagram concretizes the transportation plan and visually displays it as the train schedule, that is, the concrete plan of transforming the train flow into the subway transportation products. The vehicle operation plan stipulates the tasks of each vehicle to ensure the smooth implementation of the operation diagram (Fig. 2). 4.2

The Points of Organization Integration

The points of organization integration include the integration of the vertical layout and the integration of the lines. (1) The vertical organization integration of operation planning The vertical organization integration is to organize the independent operation plans of urban rail transport lines vertically, which is completing the integration of the coordination of each train operation plan on the basis of transportation capacity analysis. Speciﬁcally, it is to set train departure frequency, analyze the constraints of the train operation process and determine the occupied interval and station time according to the characteristics of the line, after determining the train operation routing, stop scheme and marshaling scheme in each period.

Connotation and Direction of Urban Rail Transport Integration

303

Fig. 2. The map of the core and relationship among train operation plan

(2) The compilation organization integration of operation planning The compilation organization integration of urban rail transport operation planning is to realize the supply and demand balance and beneﬁt optimization of urban rail transport network by the collaboration through operation plans on the basis of the background of urban rail transport network. The compilation organization integration aims to optimize the over-line passenger flow. Based on the analysis of the relationship among passenger flow of transfer stations in the network, it optimizes the rail transport network train arrival and departure connection aiming at achieving the goal of maximizing the efﬁciency of transferring passenger flow. In addition, as the extend of network transferring, the setting of the departure time of the ﬁrst and last train is also the research point of the compilation organization. 4.3

The Development of Organization Integration

With the development of interactive design, it is predicted that the vertical organization integration will take the change of train departure frequency as the dominant means to coordinate the contradiction between supply and demand of transport capacity. That is, on the basis of the division of the time interval, real-time data acquisition and real-time response of the passenger flow of each station are carried out, and the train departure frequency is adjusted in real-time according to the real-time passenger flow demand. As for the compilation integration, since the urban development will lead to the expansion of network, the increase of transfer nodes and the uncertainty of passenger dynamic changes will lead to the increase difﬁculty of transfer synchronization and coordination control. Therefore, the real-time allocation of buffer time and real-time adjustment of departure interval will be the trend of compilation organization integration.

5 Epilogue The Integration of urban rail transport system involves many aspects such as road network construction, hub connection, transportation organization and management system. But its core is on the basis of urban planning and passenger transportation demand, to ultimately achieve the overall efﬁciency of the passenger transport system

304

T. Wu et al.

optimal and sustainable development. This paper mainly introduces the integration of planning, connection and organization, and predicts its development trend. On the basis of this paper, the further research and exploration of related realizing technique during each period is needed. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Deng, M.Y.: Planning study on the link between track trafﬁc and other modes of trafﬁc. Planners 20(8), 76–78 (2004) 2. Ma, L.B.: Integration of urban rail transit and urban planning. Technol. Appl. 1, 22–23 (2018) 3. Ye, H.: Discussion on coordinated development of rail transit and city. J. Suzhou Univ. Sci. Technol. (2010) 4. Lin, J.: Integrated design of urban rail transit station. Railw. Investig. Surv. 3, 83–85 (2016) 5. Bian, Y.D., Yang, Y.P.: Key issues of sustainable and healthy development of urban rail transit. Urban Rapid Rail Transit. 25(2), 13–15 (2012) 6. Li, J.Y.: Study on optimization of conventional bus routes based on rail transit. Urban Rapid Rail Transit. 22(3), 10–13 (2009) 7. Qian, H.F., Huang, X.F.: Study on the Connection between urban rail transit and conventional public transport. Technol. Innov. Appl. 20, 16–17 (2012)

Innovation of Networked Railway Transportation Organization in High-Speed Railway Shaoquan Ni1,2,3(&), Feiyu Yang4, and Miaomiao Lv1,2,3 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Operation Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 School of Information Science and Technology, Southwest Jiaotong University, Chengdu 610031, China

Abstract. With the optimization and adjustment of the economic and social structure and operation mode of our country, the railways in China has entered a new era of “double network integration” with high-speed railway network and the normal speed railway network emerging together, which has led the global railway development. However, with the integration of the two networks, the problems of China’s railways have gradually emerged. Therefore, the main problems existing in railway transportation service mode and capacity resources allocation, information support, operation design and transportation organization in China’s railway are analyzed in this paper, and innovation of railway transportation organization based on regional coordinative high-speed railway transportation organization, fright transportation organization with transit period, theory and technology of dynamic transportation plan compilation is put forward. At the same time, a brief description and analysis of complete sets technology of the market oriented dynamic planning for railway transport organization is given. Keywords: Innovation of transportation organization Networking conditions

High-speed railway

Judging from the current railway situation, China’s economy and society have entered a new era when the demand of passenger transportation has become increasingly diversiﬁed. With the rapid development of science and technology and the integration of high speed railway network and normal speed railway network, a complex passenger transport system has gradually formed which is integrated into network by the normal speed railway, the intercity railway and the high-speed railway. As a result, China’s high-speed railway lines have strong correlation and dynamic influence. And the operation line of the entire road network is intertwined, complicated and interlocking. So, it is imperative to construct an innovation system of engineering technology for transportation organization that meets the needs of railway operation strategy © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 305–311, 2019. https://doi.org/10.1007/978-3-030-04582-1_35

306

S. Ni et al.

transformation to enhance the national strategy of railway services and the major needs of social and economic development capabilities and the backbone role and status of the railway in the integrated transport system. Therefore, it is high time to break through the traditional theory of high-speed railway transport organization, to realize the theory innovation under the condition of network and improve the level of railway transport organization in China.

1 Research Background As China’s economy and society enters into a new era, the high-speed railway network and normal speed railway network have been integrated and a rapid development of science and technology has been achieved. But, due to tremendous changes in internal and external environment, railway transportation industry now confronts with both opportunities and challenges. And the main problems of railways in China existing at present are as follows: In terms of transport service mode and transportation resource allocation, the high speed and normal speed railway in China adopts the mode of networked through operation and cross lines transportation for trains with multiple speed grades. However, that there is no signiﬁcant difference in the division of work between the high-speed railway and the general railway and no high-level freight logistics, together with the capacity intensity of main passageways and hubs, lead to prominent structural problems in the resource allocation [1]. In terms of operation design and transport organization, the technical method system of transport organization with capacity constraints has been established based on the long-term operational practice of China’s railways. The main problems are that there is a low degree of coordination in freight marketing, plan compilation, and dispatching command, freight trains cannot run strictly in accordance with the operation diagram, thus it is impossible to fundamentally meet the requirements of fright transit period; the difference in the function positioning of passenger transport in highspeed and normal speed railway is not signiﬁcant, the pedigree of passenger product is incomplete, and the passenger organization is not sensitive enough [1]. In terms of information support, the informatization construction of railway in China has played an important role in ﬁelds of passenger and freight marketing, transportation organization, operation safeguards, information service and so on. However, as a result of the low degree of the integration of information resources and data sharing, and business applications which is based on the concept of traditional transport organization lacking in integration, it failed to effectively support the process reengineering and innovation breakthrough of transport organization [1]. At present, the problems mentioned above have led to the inadequate performance of efﬁciency, inadaptability to market changes, the low degree of cooperation in operation safeguards, and the insufﬁcient support of information of railway in China.

Innovation of Networked Railway Transportation Organization

307

2 Solution From the perspective of industry development, the railway transportation industry in China presents a trend from production to service, and from the perspective of theory and technology development, it embodies the trend of dynamic transportation decisionmaking, integration of operation safeguards, and integration of information systems. In addition, the rapid development and mature application of transportation basic theory, big data, Internet plus, Internet of things, artiﬁcial intelligence and other technologies provide the feasibility for the innovation of railway transport organization. 2.1

To Innovate the Mode of Railway Service

In terms of service mode, the rational division of work between the high-speed railway and the normal speed railway will be studied with the guidance of “short-distance passenger transport enjoys the ﬁrst priority in high-speed railway, while in normal speed railway, priority will be given to long-distance passenger transport and freight transportation.”; the innovation of high-speed railway passenger service mode will be studied by taking “transfer takes the ﬁrst place in cross-regional passenger flow, and direct transport takes the ﬁrst place in intra-regional passenger flow.” as the guiding ideology; and the “adapting to dynamic change of the market and the requirements of the time limit, the coordination of marketing and planning, and realizing the train operation in accordance with the train diagram.” is took as guiding ideology to study the innovation of logistics service mode of freight transportation, so as to improve the comprehensive efﬁciency and service level of the railway. 2.2

To Innovate the Theory of Transportation Organization

In terms of transport organization, though the train diagram serves as the basis of operation organization, the existing compiling mechanism cannot adapt to market changes. There are still problems including the difﬁculty in guarantee the transit period in freight transportation, trains’ not operating completely in accordance with the train diagram, and insufﬁcient passenger transport sensitivity. In order to give full play to the railway efﬁciency and improve the railway service level, a market-oriented dynamic planning transportation organization model will be studied and established, in which marketing decisions is made and daily train operation diagrams is implemented according to the market demand and resource status so as to form a precise dynamic transportation plan, and innovative theory centered on dynamic transportation planning will be established to break through the technical bottleneck of railway transportation organization [2]. 2.3

To Innovate the Railway Transportation System

In terms of information technology, a multi-source and big data platform for railway transportation will be further developed ﬁrst to achieve data integration and sharing. Then, the dynamic transportation planning system will be developed in order to support the dynamic optimization of the network resources, as well as to realize the seamless

308

S. Ni et al.

link between demand, marketing and planning. On this basis, an intelligent railway transportation system with dynamic transportation planning as the core will be developed to support the railway’s adaptation to the dynamic changes of the market, and to realize the synergy, agility and intelligence of the railway transportation organization and improve the railway service level [3].

3 Innovation of Theory and Technology of Transportation Organization 3.1

Theory and Technology of Regional Cooperative Transportation Organization for High Speed Railway

In order to optimize the high-speed railway service structure under the networked conditions and to highlight the dominant position of the high-speed railway in the medium and short distance passenger transport market, the existing high-speed railway cross-line passenger flow organization model with direct transportation in the ﬁrst place is broke through to realize the reconstruction of the theory of high-speed railway transportation organization with the mode innovation of “transfer of cross-region passenger flow and direct transport of intra-region passenger flow” as the core. In addition, key issues including the reasonable operation distance of high-speed train, regional network division method for high-speed railway, transfer node selection of cross-line passenger flow, the innovative theory and technology of regional cooperative transportation plan compilation for high-speed railway, and the proposal of coordinated and matching methods and technology for the improvement of transportation organization and service level in high-speed railway including ticket sales, transfer organizations, dispatching command, information services, etc. is studied. 3.2

Theory and Technology of Transportation Organization of Freight Transit Period

First, the precise transportation of the whole process from order to delivery is studied and the innovation mechanism of freight organization is put forward, and the freight business and organization process is reconstructed, so as to improve the response speed of transportation service towards the market demand and build a dynamic and adaptable railway freight production system. Second, the key issues such as collaborative optimization of dynamic transport planning and transportation resources, and combination of the train marshalling plan with the streamline of train diagram ought to be studied. Third, it is necessary to establish a dynamic planning freight organization innovation theory which is with the core of the whole network dynamic transportation planning and with the goal of ensuring the freight transit period, driving in accordance with train diagram, to achieve precise transportation.

Innovation of Networked Railway Transportation Organization

3.3

309

Theory and Technology of Dynamic Transportation Planning

With the starting point of meeting the demand of dynamic market, a railway-wide dynamic transportation planning system that responds quickly to customer needs is built based on the basic operation diagram, to strengthen the accuracy and feasibility of the plan and to ensure the trains’ driving in accordance with the diagram. Taking the customer as the center, the integrated planning technology with the integration of production, transportation and marketing and the integrative assembling, evacuating and transporting is studied; and the theory and technology of cooperation compilation between train operation plan and train operation diagram is studied, the cooperative compilation theory between train operation plan and locomotives and rolling stock, crew scheduling, station operation and comprehensive maintenance is established. Based on optimization theories and methods such as deep learning, research on the intelligent optimization technology of planning is conducted, and the innovative theory and technical system of railway transportation planning under the conditions of dynamic transportation demand and transportation resources, such as passenger and cargo marketing, trafﬁc organization and train operation organization is constructed.

4 Complete Sets of Technology for Dynamic Planning Market-Oriented Railway Transport Organization 4.1

Problem Statement

China’s railways have entered the new era when high speed railway network and the normal speed railway network merge together, which has led the world’s railway development. And the tension in railway transport capacity has been effectively alleviated. But, at the same time, the rapid development of society and economy has continuously increased the requirements for diversiﬁcation, high quality and timeliness of railway passenger and freight services. Although the strategic concept of “full stream of service and production” is put forward by the railway, the theory and technology of railway transport organization did not achieve innovative breakthroughs, and could not fundamentally meet the needs of market dynamics, which results in insufﬁcient social functions of railways. Therefore, to establish a new mechanism for railway transportation organization that comprehensively, quickly and accurately responds to market dynamic changes fundamentally and enhance the ability of railways to adapt to market dynamics and support the transformation of railway operations from production to service is the fundamental way for the development of railway transportation. 4.2

Key Difﬁculties and Challenges in the Future

(1) Theoretical innovation: Based on the research on the precise transportation and freight organization innovation mechanism of the whole process from order to delivery, the freight business and organizational process are reconstructed to improve the response speed of transportation service to market demand, and further construct a dynamic and adaptable railway transportation production

310

S. Ni et al.

organization system. Finally, a dynamic planning innovation theory of railway organization with the goal of ensuring the fright transit period, trains’ driving in accordance with the operation diagram and precise transportation, and with the core of the achievement of in the whole network and the whole process is established. (2) Technological innovation: Through the establishment of a complete set of key technical methods that are compatible with the transportation organization innovation theory in the marketing, product design, planning, dispatching and commanding, as well as the development of intelligent railway transportation with the whole network dynamic train operation diagram as the core, the integration of railway transportation demand, marketing decision, transportation planning, ﬁxed equipment and facilities monitoring, mobile equipment history and dynamic management, and personnel dynamic management can be realized. 4.3

Signiﬁcance

In-depth study of the intelligent railway transportation system and technology based on the dynamic railway train operation diagram, and the construction of technical support means for modern railway transportation organization is the fundamental guarantee for the modernization of railway transportation organization, and will also be the successful model of integration of the information science and railway transportation engineering organic, which will also promote the in-depth research and application of information technology and modern artiﬁcial intelligence technology in the ﬁeld of railway transportation, and lead the deep research and application of information technology and intelligent technology in the ﬁeld of planning and production organization. Moreover, the intelligent transportation system of the intelligent railway with the adaptation of market change, high integration of the whole road network and the development of dynamic train diagram as the core is to break through the technical bottleneck of the development of railway transportation and logistics, and to fundamentally innovate the theory and technical means of railway transportation organization, and further improve the ability and level of the railway service for society and economy. 4.4

Market Oriented Dynamic Railway Transportation Planning System

On the one hand, the existing railway passenger transport organizations have poor sensitivity and cannot carry out precise transportation organization according to market dynamic changes. On the other hand, with the development of globalization and modern logistics industry, the demand for freight transportation is more time-sensitive, thus the construction of railway transportation organization innovation theory with the dynamic train operation diagram as the core and the development of the dynamic train operation map compilation system in the whole network and the whole process serves as the key to the competitiveness of the transportation market. Moreover, based on the big data platform of railway transportation, the collection of passenger flow, cargo flow, transportation capacity resources and train running state can be realized, and

Innovation of Networked Railway Transportation Organization

311

based on market dynamics, the coordinative compilation of freight planning, train operation plan, locomotive and rolling stock application and maintenance plan, crew plan, station operation plan and comprehensive maintenance plan can be also realized.

5 Conclusion Through analyzing the socio-economic environment of railway transportation development in China at present and the problems existing in the current railway development, combined with the strong correlation and dynamic influence of high-speed railway lines and the intertwined, complicated and interlocking characteristics of the operation line in the entire road network, it is proposed to establish a dynamic planning-type transportation organization theory and technology with the whole network dynamic transportation planning as the core and fundamental solution to railway freight transportation’s responding to market changes, which is conducive to improving the sensitivity of railway passenger transportation and is groundbreaking in the theory and technology of railway transportation organization. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Shao, C., Lv, M., Zou, C., Ni, S.: The plan of the stop station of the Beijing Shanghai highspeed railway based on the normalization of the train operation diagram. Railw. Transp. Econ. (7), 1–6 (2018) 2. Xu, Z., Ni, S., Zhu, Q.: Study on the organization mode of freight transport in Monghua Railway based on fright transit period. J. Transp. Eng. Inf. 16(1), 131–137 (2018) 3. Ni, S., Zhao, C., Zhuang, H., Lv, H.: Principle and Method of Computer-Aided Train Operation Diagram Compilation. Southwest Jiaotong University Press, Chengdu (2017)

Discussion on the Application of Big Data in Rail Transit Organization Guobao Du1, Xuepeng Zhang1, and Shaoquan Ni2,3,4(&) 1

Transport Ofﬁce, China Railway Chengdu Bureau Group Co., Ltd, Chengdu 610031, China 2 School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 3 National Railway Train Operation Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 4 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China

Abstract. At present, the Internet technology has been widely used in people’s life. In this “information” era, the application of big data technology in the ﬁeld of rail transport is particularly necessary. In view of this, the industry background and the theoretical basis of application is analyzed ﬁrst in this paper, and then the application of big data in rail transportation ﬁeld on the basis of railway passenger and freight marketing, railway operation and dispatching, railway passenger and freight service is discussed. At the same time, according to the dynamic planning of the railway transport organization system, the transport organization theory innovation and information system innovation are briefly analyzed. Keywords: Rail transport

Application of big data

In recent years, with the rapid development and mature application of Internet technology, the data that people have contacted and used has exploded, and the analysis and processing of big data has gradually been applied to all walks of life. For rail transportation industry, it shoulders the arduous task of railway passenger and freight transportation. In the context of “big data” era, the application of big data in the ﬁeld of rail transit is briefly discussed in this paper.

1 Industry Background and Basic Overview 1.1

Industry Background

Accelerating the development of rail transit has become a national strategy. The 19th National Congress of the Communist Party of China clearly put forward the ambitious goal of building a powerful transportation country. The development plan of the modern comprehensive transportation system in the “13th Five-Year Plan”, the outline of the medium and long-term scientiﬁc and technological development plan, © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 312–318, 2019. https://doi.org/10.1007/978-3-030-04582-1_36

Discussion on the Application of Big Data in Rail Transit Organization

313

the medium and long-term railway network plan and the “13th Five-Year” development plan of the railway clearly put forward the need to speed up the development of rail transit. There will be much to beneﬁt from the research on intelligent transportation of rail transportation. With the rapid development of China’s economy and the acceleration of urban agglomeration, rail transit construction is in full swing, the scale of high-speed railway and urban rail transit is growing rapidly, giving full play to the social functions of rail transit, and improving the level of rail transit operations in China is a matter of national economy and people’s livelihood, which is in urgent need of a large number of rail transit intelligent transportation technology with independent intellectual property rights as a support [1]. 1.2

Application Basis

Based on the big data application method proposed by McKinsey, it’s not difﬁcult for us to acknowledge the following things. Firstly, it will be easier for stakeholders to get the value of big data by creating transparency. Secondly, requirements can be discovered by enabling experiments, which lead to variability and performance improvement; Thirdly, by subdividing the crowd, flexible actions could be taken to get closer to real-time analysis and provide precise services. Finally, based on big data analysis, automation algorithm can be adopted to replace or assist human decision making. 1.3

Overview of the Application of Big Data on the Railway at Home and Abroad

(1) Passenger and freight transportation: The use of big data to accurately predict passenger flow and freight flow, grasp the market demand and change trends, and provide personalized services and precision marketing is conductive to improve the level of railway service and transport revenue. (2) Equipment maintenance: Using big data to master equipment status, predicting the development trend of fault has contributed to the transformation of “plan repair” to “state repair”, and the cooperative linkage between transport organization and operation and maintenance of the equipment. (3) Security: Multi-factor impact correlation analysis based on big data for transportation security is conducted to achieve security warning and risk management. (4) Business development: The business model of value-added utilization of railway big data resources has been established, and a new format for the development of the railway industry has been formed. (5) Intelligent railway construction: Based on big data technology and railway BIM technology, the digital real-time online operation environment of the railway is constructed, which provides digital support for the integration of railway construction, operation, dispatch, emergency and disaster prevention.

314

G. Du et al.

2 Application of Big Data in Railway Passenger and Freight Marketing 2.1

Passenger Flow Investigation

Mobile localization technology of the smartphone can be used to get residents’ daily trip trajectory. Through the mobile localization technology, people’s spatial and temporal information can be easily got, which can avoid the shortcomings of the traditional questionnaire survey and obtain continuous OD data for a period of time. 2.2

Customer Relationship Management

Through the big data technology, the behavior information of passengers and owners can be integrated, and customer travel habits and delivery habits can be mined, and deeply analysis of customer loyalty and satisfaction and customer value is conducted to predict customer’s long-term growth potential, assess customer’s loss risk and make early warning. Apart from this, classiﬁed marketing can be carried out to customers to screen out the core target customers and push the information to the customers accurately through the active marketing, service marketing, price marketing and other different ways. 2.3

Passenger Flow Forecasting

Through dynamic acquisition of data in ticketing system and automatic ticket system, as well as weather, large activity and other external data, massive raw information can be collected, and through the analysis of passenger flow characteristics, the spatial and temporal distribution of passenger flow can be analyzed in real time to explore the law of passenger flow and predict the trend of future passenger flow, so as to provide necessary reference information for operation management. 2.4

Freight Demand Analysis

Through the analysis of the data of the electronic commerce system of railway freight, the change rules of the demand, category and direction of freight transportation in different seasons can be accurately grasps. 2.5

Precision Freight Marketing

Through the transaction data of the customer on the e-commerce website, as well as the data of enterprise production, circulation, operation, ﬁnance, sales, customers, and the related industrial chain in the marketing process, the current demand of the enterprise and the development of the external environment of the enterprise can be better grasped, and the future situation of the enterprise can be predicted; clustering and subdivision of the customers can better allocate resources and improve services. Through effective information collection and data mining and analysis, marketing strategies can be formulated in a targeted manner.

Discussion on the Application of Big Data in Rail Transit Organization

315

3 Application of Big Data in Railway Operation and Dispatching 3.1

Adjustment of Product Design and Planning

Through large data technology, the changing law of peak season and off-season can be grasped and rolling identiﬁcation of short-term situation can be achieved to dynamically adjust train operation plan and transportation plan, innovate and design marketable passenger and freight products, intelligently push product information needed by passengers and cargo owners, so as to make it more suitable for the actual needs of passengers and cargo owners and save cost and improve economic performance. 3.2

Analysis of Train Operation and Dispatching Command

By using big data technology, the basic train diagram is compared with the actual train operation performance, the train operation law including inertia delay is analyzed, and the reasons for the train operation deviation from the planned train diagram are analyzed with the principle of relevance, which can provide basis for optimizing the train operation diagram and improving the reliability of railway transportation services. 3.3

Assistant Decision Making for Emergency

According to the dynamics of materials and equipment, an emergency plan is formulated, and an emergency response mechanism is automatically activated to allocate human and material resources. The personnel and equipment on the whole line should be monitored to analyze the task execution and the status of spare parts of staffs at each site. Through the deep mining of the historical frequency of equipment and facilities failure and the flow of data, a reasonable emergency resource allocation scheme is determined to deploy and allocate resources rationally. 3.4

Analysis of Passenger Travel Characteristics

Through the big data platform, the passenger travel habits can be timely mastered, the early and late peak hours can be predicted to adjust the train operation and determine the flow limit scheme, which is conducive to the planning adjustment and dispatching command.

4 Application of Big Data in Railway Passenger and Freight Service 4.1

Urban Rail Transit Information Service

Through real-time monitoring of passenger flow in various sections of rail transit lines and passenger flow density and operation status in special locations and displaying the operation congestion degree of each line, passengers are convenient to adjust the plan

316

G. Du et al.

of taking metro and avoid the congestion and fault areas. It can also provide travel information service through station platform, train display screen, self-service inquiry screen, metro website and other main carriers, which is aimed at different periods, such as waiting time, riding time and time before going out. 4.2

Railway Passenger and Freight Service

(1) Information service for passengers in the station Through the collection of the status of the train at the station’s being on time or late, and the service status of the key service facilities such as ticket selling equipment, ticket checking passage and elevator, and by using the large data analysis method to analyze and predict the passenger travel characteristics and trends, the passenger demand can be accurately grasped, and effective passenger transport organization measures can be formulated to provide more accurate and individual extension service for passengers. (2) Passenger service Big data can also be used to enhance the interaction and experience of passengers. Through the development of train real-time positioning system, passenger trains can be tracked and located, and train real-time positioning information is displayed based on the map, which makes the dynamic of transfer trains clear at a glance. And by mining passengers’ preferences, the type of meals provided by each train can be set dynamically according to the passengers’ identity information, and appropriate commodity marketing plans can be switched in time to provide personalized services. (3) Freight service Through the development of the train real-time positioning and freight tracking system, the freight can be tracked and located, and the real-time positioning information of the train can be displayed based on the map, which can provide the shippers and logistics companies with passive and active information services.

5 Dynamic Planning Railway Transportation Organization System 5.1

Theoretical Innovation of Transportation Organization

In terms of transport organization, though the train diagram serves as the basis of operation organization, the existing compiling mechanism cannot adapt to market changes. There are still problems including the difﬁculty in guarantee the transit period in freight transportation, trains’ not operating completely in accordance with the train diagram, and insufﬁcient passenger transport sensitivity. In order to give full play to the railway efﬁciency and improve the railway service level, a market-oriented dynamic planning transportation organization model will be studied and established, in which marketing decisions is made and daily train operation diagrams is implemented according to the market demand and resource status so as to form a precise dynamic

Discussion on the Application of Big Data in Rail Transit Organization

317

transportation plan, and innovative theory centered on dynamic transportation planning will be established to break through the technical bottleneck of railway transportation organization. 5.2

Information System Innovation

In the aspect of informationization, it is necessary to develop an intelligent railway transportation system with dynamic transportation planning as the core on the basis of large data platform, so as to realize the integration and informatization of demand collection, marketing decision-making, transportation planning, ﬁxed equipment and facilities monitoring, mobile equipment resume and dynamic management, personnel dynamic management, dispatching command and train transportation, and automatic train control. In addition, on the basis of dynamic monitoring and management of transport capacity and resources, a comprehensive transport plan is formulated based on dynamic demand, and the dispatching command is adjusted to realize the integration and intellectualization of railway transport organization. That is to achieve the informationization, cooperation and intelligence of the transportation organization through comprehensive information collection, so as to support the coordination, agility and intelligence of the whole transportation organization of the railway network, and improve level of the railway service.

6 Conclusion Based on the background of the times, the development background of the rail transportation industry and the application basis of data technology is analyzed in this paper. In railway passenger and freight marketing, the application of big data technology in passenger flow investigation, customer relationship management, passenger flow forecasting, freight demand analysis and precision freight marketing is expounded. In railway operation and dispatching, the application of big data technology in the aspects of the adjustment of product design and planning, the analysis of train operation dispatching and command, assistant decision making for emergency. And based on the information service of urban rail transit and passenger and freight service, the application of big data technology in passenger and cargo service is analyzed as well. Finally, based on the dynamic planning type railway transportation organization system, the innovation of transportation organization theory and information system is briefly analyzed. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015,2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project(2017-RK00-00028-ZF, 2017-RK0000378-ZF)and the Fundamental Research Funds for the Central Universities(2682017CX022, 2682017CX018).

318

G. Du et al.

References 1. Dai, M.: Thoughts on the application of big data technology in China’s railways. Railw. Transp. Econ. 3, 23–26 (2014) 2. Zheng, J.: Research and application status of big data in railway abroad. China Railw. 2, 54– 62 (2018) 3. Wang, T.: Research and practice on top-level design of the application of big data in China’s railway. China Railw. 1, 8–16 (2017)

Long and Short Routing Mode of Nanjing Railway Transit Line 3 Rong Kuang1(&), Wenxian Wang2, and Tingting Wu1 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu, China [email protected], [email protected] 2 School of Railway Transit, Wuyi University, Jiangmen, China [email protected]

Abstract. Based on the analysis of main factors of Nanjing railway transit line 3, including the function location, station distribution and passenger flow characteristics, the article proposes ﬁve alternative routing operation schemes. Taking service quality, turn-back ability, safety and flexibility of the operation organization, operation cost and long-term development ability into consideration, the evaluation system of operation scheme of Nanjing railway transit line 3 is constructed. With a comparative study of the ﬁve schemes through analytic hierarchy process, the best operation scheme of long and short routing mode of Nanjing railway transit line 3 is obtained. Keywords: Urban rail transit Analytic hierarchy process

Long and short routing

1 Introduction With the acceleration of urbanization process, the city scale is expanding and residents are traveling more and more frequently, there is no doubt that urban transportation faces the grim situation. Thanks to its unique advantages such as large volume, fast speed, low energy consumption and strong punctuality, urban rail transit has become the most effective travel tool to alleviate urban congestion problem as well as played the key role in modern urban public transport. The routing operation scheme is essential to daily operation. It speciﬁes the running section, turn-back station, types and number of passenger trains. The main purpose of setting up reasonable trafﬁc is to satisfy the passenger transportation demand, facilitate operation, distribute transportation capacity rationally and save vehicle equipment [1]. Based on the analysis of conditions of Nanjing railway transit line 3, including brief introduction, function position and passenger flow indicator, this paper proposes ﬁve schemes and make comparisons, aiming to select the best operation scheme of long and short routing mode of Nanjing railway transit line 3.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 319–326, 2019. https://doi.org/10.1007/978-3-030-04582-1_37

320

R. Kuang et al.

2 Train Operation Routing Mode Establishing train routing is an effective way to solve the unbalanced passenger flow in different section of the line. Routing operation schemes is mainly based on the prediction of maximum transect volume during peak hours in the year of planning. Its compiling should consider following factors: ﬁrstly, according to the idea of train determined by flow [2]. dividing operation periods and sections to count passenger flow volume. Secondly, measuring the capacity of turn-back section to avoid the waste of transport energy in the initial stage of construction. Thirdly, Consider the needs of forward passenger flow. The rationality of the routing mode is reflected in two aspects. For one, ensure high quality of service. For another, focus on the operability of operations management and the flexibility of the operating organization. The basic forms of rail transit routing mode include single route, long and short route, sectional running route, staggered running route and so on [3]. Single route runs through the entire line from the start station to the end station. With the increase of the length of lines and the continuous development of network construction, it can not meet the requirement of passenger flow. Many cities have tried to adopt long and short routing mode in actual operation. When facing emergencies, including sudden large passenger flow and limited number of vehicles, it plays an important role [4]. Line 3 is second north-south passenger flow backbone lines in Nanjing. The route starts at the Linchang Station and ends at the Mozhoudong road Station, connecting six districts in series. Total length is 44.9 km. There are 29 stations in total.

3 Long and Short Routing Mode of Nanjing Railway Transit Line 3 3.1

Passenger Flow Demand Analysis

The operation of Nanjing railway transit line 3 has greatly improved the connectivity and transfer efﬁciency of Nanjing rail transit network. The passenger trafﬁc volume reached 407 thousand on the opening day. By December 2017, the maximum daily passenger trafﬁc volume reached 886 thousand. Combined with the population distribution in Nanjing, the trafﬁc pressure on the North-South main road is much greater than that in the east-west trafﬁc arteries. The maximum cross section passenger flow in the future peak period is evaluated and forecasted. With the gradual completion of the network structure, line 3 needs to undertake higher passenger flow density. There is a signiﬁcant imbalance in time, direction and cross section of the rail transit passenger flow. Time imbalance reflects that morning and evening peak passenger flow is signiﬁcantly higher than other period. Even on the same route, different direction has different volume, which belongs to the direction imbalance. Difference of passenger volume between sites shows the unbalanced cross section. The cross-section passenger flow of line 3 presents two asymmetric spindle-shaped distributions. The following 7 stations have large passenger flow volume: Liuzhoudong Road Station and Tianruncheng Station, Fuzimiao Station, Jimingsi Station, Nanjing Station,

Long and Short Routing Mode of Nanjing Railway Transit Line 3

321

Nanjing South Station and Daxinggong Station. The volume of line 3 is mainly concentrated in the middle section from Taifeng Road station to Nanjing South Station. 3.2

Long and Short Routing Schemes

The differences between train routing schemes are reflected on the operation modes and layout of turn-back station. In order to facilitate operation management as well as make full use of station equipment, the pre-designed routing scheme should create conditions for the implementation of the post-designed routing scheme. The latter should not only take the utilization of equipment into consideration, but also be focus on continuity of operation management [5]. Aim to arrange the transportation capacity reasonably, ﬁve alternatives are proposed, shown from Figs. 1, 2, 3, 4 and 5.

Linchang Station

Nanjing Station

Mozhoudong Road Station

Linchang Station

Fig. 1. Schematic diagram of scheme 1

Linchang Station

Daming Road Station

Fig. 3. Schematic diagram of scheme 3

Linchang Station

Fig. 2. Schematic diagram of scheme 2

Linchang Taifeng Road Station Station

Mozhoudong Road Station

Nanjing Station

Shengtaixi Road Mozhoudong Road Station Station

Nanjing South Mozhoudong Road Station Station

Fig. 4. Schematic diagram of scheme 4

Mozhoudong Road Station

Fig. 5. Schematic diagram of scheme 5

Five alternatives all use long and short routing mode. On the basis of actual passenger flow demand, organizing different marshalling and different number of vehicles running in each section to improve the utilization rate of transport capacity. 3.3

Schemes Analysis

Taking service quality, turn-back ability, safety and flexibility of the operation organization, operation cost and long-term development ability into consideration, the above alternatives are compared and analyzed in order to get the best solution [6].

322

R. Kuang et al.

(1) Service quality The main basis for judging the quality of operation is whether it matches the cross-section passenger flow. Evaluation of passenger for service quality have the following influencing factors. Waiting time is related to the length of short routing. During peak hours, the cross-section passenger volume has a sharply increase, so the more station it contains, the better it can meet the passenger demand as well as relieve the crowding. The transfer convenience mainly refers to the conﬁguration of turn-back stations on the short routing. In addition, short routing trains need to be cleared when they arrive at the terminal, as a result the train directness is also one of the important factors. The speciﬁc analysis of the passenger service quality of each routing scheme is shown in Table 1. The score represents the service quality. 3 points represent good level, 2 points represent the general level, and 1 points represent poor level.

Table 1. Analysis of the influence factors of passenger service quality Scheme Number Matching degree with cross-section passenger flow Waiting time Platform congestion degree during peak hour Train compartment congestion during peak hour Transfer convenience Train directness

No. 1 No. 2 No. 3 No. 4 No. 5 2 3 3 3 3 2 3 3 3 3 1 3 2 3 2 1 3 2 3 2 3 2 2 3 3 2 3 3 3 3

(2) Turn-back ability Organizing trains for short routing can adjust the situation of inadequate transport capacity in busy sections, so as to solve the problem of unbalanced transport capacity. It can also provide new ideas for driving adjustment and transport operation when temporary accidents happen. The alignment of the train needs to be carried out by means of crossover. Therefore, it is necessary to ensure the quantity and location of stations with turnouts as well as the rationality of crossover arrangement. There are 8 stations that have turn-back conditions for organizing trains along line 3. Linchang Station connects the Linchang Car depot in the up direction. Its arrangement of the crossover and turnout could realize the trains turn-back in both direction, same as Mozhoudong Road Station. The Southeast University Jiulong Station could carry out alignment in the up direction. Others could carry out the alignment in the down direction. Above analysis, scheme 1, scheme 2 and scheme 5 have obvious advantages. (3) Safety and flexibility of the operation organization Conflicting route with long routing trains is forbidden in actual operation. Therefore, the choice of the start and terminal stations of the short route is particularly important. In the scheme1, when the train turns back at Nanjing Station, which is the terminal of the short routing, there may be a route conflict with the long routing train. It is inconsistent with safety principles. The same problem

Long and Short Routing Mode of Nanjing Railway Transit Line 3

323

exists in scheme 2 and 3. In the last two schemes, conflicting route may occur at both starting and terminal station. Some factors are important to judge whether the train operation of short routing scheme is flexible, which contains length of short routing line, the layout of crossover and storage siding at the turn-back station. Line length of scheme 1 is shortest and scheme 2 is longest, others are similar. From the perspective of layout of turn-back facilities, scheme 1, 2 and 5 could realize the turn-back at both start station and terminal station, offering flexible operation organization. On the whole, scheme 1 is best one. (4) Operation cost Operation cost is affected by two factors: the length of the line and the conﬁguration of the crossover. Urban rail transit needs to rely on government subsidies to maintain operation. The longer the line is, the more costs are. Stations without crossover need to pay for the construction of turnouts, increasing cost virtually. Five schemes choose to operate long routing train from the start to the terminal. However, the length of the short routing and the layout of crossover exist difference. Based on the analysis of two aspects, the cost of scheme 3 is the highest and the cost of scheme 1 is the lowest. (5) Long-term development ability The continuity of train routing should be considered in the long-term development of rail transit. The design of rail transit line plan should be combined with the future development plan of the city. It can make greater use of the beneﬁts of rail transit network and build a comprehensive and efﬁcient transportation system. Line 3 starts from Mozhoudong Road Station in the north and goes south to Moling Street Station. The whole line is laid underground, which is 3.3 km long in total. In the ﬁve schemes mentioned above, the continuity of scheme ﬁve is the best.

4 Comparison of Alternatives Long and Short Routing Schemes Aim to select optimal long and short routing schemes of Nanjing railway transit line 3. The paper constructs the evaluation system according to the purpose and object of the study and uses AHP to sort different schemes. 4.1

Constructing Hierarchical Structure Model

The target layer contains 1 factors: the optimal routing scheme of Nanjing railway transit line 3. The criteria layer contains 5 factors: quality of service B1 , turn-back ability B2 , safety and flexibility of the operation organization B3 , operation cost B4 , and long-term development ability B5 . The scheme layer contains ﬁve alternatives. The corresponding AHP structure model is established, as shown in Fig. 6.

324

R. Kuang et al.

Fig. 6. Hierarchical analytical structure model of line 3 routing schemes

4.2

Estimate Index Weights and Calculate the Composite Weight of Each Layer

In order to ensure the authenticity and accuracy of the data, this paper collects the data through the combination of expert survey and rail transit passenger questionnaire. The evaluation factor judgment matrix A of the routing scheme is as follows. 2

1 6 1=3 6 A¼6 6 1=2 4 1=4 1=5

3 1 2 1=2 1=3

2 1=2 1 1=3 1=4

4 2 3 1 1=2

3 5 37 7 47 7 25 1

Next, the weights are obtained according to the square root method, the weight coefﬁcient are shown in Tables 2 and 3. Table 2. Weight coefﬁcient of criterion layer judgment matrix W2 W3 W4 W5 kmax CI CR W1 0.417 0.160 0.263 0.098 0.062 5.055 0.0138 0.012

Table 3. Weight coefﬁcient of scheme layer judgment matrix W1 W2 W3 W4 W5

B1 0.060 0.331 0.417 0.417 0.139

B2 0.322 0.252 0.263 0.263 0.139

B3 0.099 0.123 0.098 0.062 0.265

B4 0.342 0.075 0.062 0.160 0.265

B5 0.177 0.219 0.160 0.098 0.192

kmax 5.022 5.052 5.067 5.067 0.466

CI 0.005 0.013 0.017 0.017 0

CR 0.005 0.012 0.015 0.015 0

The above judgment matrix passed the one-time test in all four methods, CR \0:10.

Long and Short Routing Mode of Nanjing Railway Transit Line 3

325

According to the weighting coefﬁcients of each criterion layer and scheme layer, the total weights of the ﬁve alternatives are calculated as shown in Table 4. Table 4. Total weight of the ﬁve alternative Schemes

W1

W2

W3

W4

W5

B1

0.417

0.060

0.322

0.099

0.342

0.177

B2

0.160

0.331

0.252

0.123

0.075

0.219

B3

0.263

0.417

0.263

0.098

0.062

0.160

B4

0.098

0.417

0.263

0.062

0.160

0.098

B5

0.062

Index

Total sort weight

0.139

0.139

0.265

0.265

0.192

0.237

0.278

0.109

0.203

0.173

As can be seen from the above table, scheme 2 is the optimal scheme. This result is consistent with the scheme adopted by Line 3 in actual operation. Set up the long routing from Linchang station to the Mozhoudong road station, set up the short routing from Linchang station to the west of Shengtai road station.

5 Epilogue Reasonable selection of rail transit routing can make the transportation capacity evenly distributed. Based on the analysis of main factors of Nanjing railway transit line 3, the article proposes ﬁve alternative routing operation schemes. Taking several important factors into consideration, the evaluation system of operation scheme of Nanjing railway transit line 3 is constructed. With a comparative study of the ﬁve schemes through analytic hierarchy process, the best operation scheme of long and short routing mode of Nanjing railway transit line 3 is obtained. This result is consistent with the ﬁnal scheme adopted by Line 3, which veriﬁes the feasibility of the model and algorithm in this paper. It could provides a reference for the subway which has not yet opened in the selection of the mode of operation. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015,2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

326

R. Kuang et al.

References 1. Li, H.: Long and Short Routing Model of Shanghai Railway Transit Line 9. Urban Mass Transit, 2012:026 (in Chinese) 2. Li, J.: Analysis of typical intersection modes of Urban Rail Transit. Railw. Transp. Econ. 31 (10), 54–58 (2009). (in Chinese) 3. Xu, R.: Optimization of routing mode on regional express rail. Study Urban Rail Transit 5, 36 (2006). (IN CHINESE) 4. Gu, B.: Operation scheme analysis of long rail transit line. Modern Urban Rail Transit Syst. 3, 36 (2005). (in Chinese) 5. Sun, Y.: Analysis of operation mode of modern metro system in Nanjing Metro Line 1. Trafﬁc World 2008(14), 122. (in Chinese) 6. Jin, Luo: Research on operation mode of Guangzhou Metro Line 4[J]. World Rail Transit 6, 42 (2008). (in Chinese)

Transfer Optimization of Multi Standard Regional Rail Transit Zhou Xia1,2, Zhang Peng1,2, Lv Hongxia1,2(&), and Ni Shaoquan1,2 1

2

School of Transportation and Logistics, Southwest Jiao-Tong University, Chengdu, Sichuan 610031, China [email protected] National Railway Train Diagram Research and Training Center, Southwest Jiao-Tong University, Chengdu 610031, China

Abstract. This paper briefly analyzes the characteristics of two new rail transit modes, namely, City area and city neighborhood, which forms a multi-system regional rail transit system by cooperating with national railways and urban rail transit. From the angle of transfer coordination, aiming at minimizing the waiting time of passengers, this paper optimizes the arrival and departure time of trains at transfer stations, and constructs an optimization model of transfer and connection between urban rail transit and multi-system regional rail transit. Fixed the national railway train plan and adjusted the arrival and departure times of urban rail transit trains to reduce the impact on the overall plan. Case studies show that the model can reduce passenger waiting time to a certain extent, and provide a reference for the coordinated development of multi-mode rail transit. Keywords: Multimode Regional rail transit City adjacent line and city line Transfer connection

1 Introduction With the development of urbanization, the scale of cities is expanding. The single system of urban rail transit has been unable to meet the growing travel needs of people. Two new types of urban rail transit, namely, adjacent urban rail transit and urban rail transit, have emerged. They are transferring and connecting with national railways and urban rail transit, forming a multi-system regional rail transit system to adapt to social development. It is one of the directions that how to draw up the operation plan of various rail transit systems from the road network level, realize the passengers’ good transfer and continuation, and optimize the arrival and departure time of various rail transit systems at transfer stations. With the development of urban rail transit network, there are many studies on the optimization of urban rail transit network operation. References [1, 2, 3] consider the effect of train delays, and from the passenger’s point of view, to minimize the waiting time of transfer station, establish the optimal model of arrival and departure time of transfer station train. Documents [4, 5] study the departure time of the ﬁrst and last train in the network by combining the arrival and departure sequence of trains within the transfer station with the coordination between network transfer nodes. Document [6] from the perspective of transport capacity matching, the ﬁrst train departure time and © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 327–334, 2019. https://doi.org/10.1007/978-3-030-04582-1_38

328

Z. Xia et al.

train running interval are studied. Reasonable transfer and connection should be used not only in urban rail transit, but also in multi-system rail transit. At present, there is relatively little research on this aspect. Based on the research results of transfer and connection optimization of urban rail transit network operation, this paper constructs a multi-system rail transit train connection optimization model to provide direction for regional rail transit train operation planning.

2 Analysis on Transfer Connection of Multi Standard Rail Transit 2.1

Multi Standard Regional Rail Transit System

The metropolitan and adjacent lines are a new type of public transit passenger rail transit system between the trunk lines and urban rail transit, serving the inner cities of the urban agglomeration and the central cities or satellite cities. Its passenger flow components mainly include commuter passenger flow and ofﬁcial passenger flow as well as entertainment, family visits, personal affairs and other passenger flow, with the characteristics of high speed and large capacity. One end connects the city’s main satellite city, the city’s deputy center and the diplomatic hub, and the other end terminates in the city center. Deep into the city center area, extending the existing subway-based rail transit services, urban rail transit to do a good supplement. The station located at the comprehensive transportation hub station coordinates with existing railways and high-speed railways to improve the efﬁciency of the whole rail transit network. Municipal and adjacent lines closely cooperate with urban rail transit and state-owned railways, forming a multisystem regional rail transit system, and jointly promoting the rational division of labor and coordinated development of large, medium and small cities and towns. 2.2

Multi Standard Rail Transit Contact Process

It is necessary to satisfy the certain time relationship to realize the rail transit connection. Taking one-way transfer as an example, a Metropolitan line1 is set up to transfer to a transfer station e with a metro line2. Passengers transfer from the upstream direction of the metro line to the downstream direction of the Metro line. At this time, the Metropolitan line train acts as a transport train i and the Metro Line Train acts as a connecting train j. The transfer connection diagram is shown in Fig. 1.

Fig. 1. Schematic diagram of transfer waiting time

Transfer Optimization of Multi Standard Regional Rail Transit

329

The horizontal axis represents the time, the arrow pointing and deviating from the time axis respectively indicates the arrival and departure of the train. A1;i indicates the time point of the train i arriving at the transfer station e, L2;j indicates the time point of the train arriving at the transfer station, and D2;j indicates the stop time of the train station. t12 Indicates the time when the passengers of the line1 arrive at the line2 through the transfer corridor and the waiting time of the passengers at the platform. The problem to be solved in the transfer of multi-system rail transit is to make the waiting time of passengers as small as possible. The problem to be solved in the transfer of multi-system rail transit is to make the waiting time of passengers as small as possible. Of course, not waiting for the shorter, the better. When the waiting time is relatively small, passengers may face the just miss. Therefore, wi;j should be kept within reasonable limits for passengers to travel comfortably and quickly.

3 Multimodal Regional Rail Transit Optimization Model 3.1

Problem Assumptions

For the convenience of research, the following assumptions are made: (1) The independent operation of the lines of each standard will not affect each other. (2) For the same line, there is no transfer relationship between the upper and lower directions. (3) During the coordination period, the various lines of each standard are evenly spaced. (4) During the coordination period, the transport capacity is sufﬁcient, and the transfer passengers can take the ﬁrst train to leave the platform. 3.2

Model Establishment

Within the coordination period ½ta tb , for a transfer station, Set up rail transit standard collection s 2 ðA National Railway, B City neighbouring line, C City area line, D metroÞ, each type has ns rail transit lines. Separate the two directions of uplink and downlink separately. Use Rsij to indicate the j direction of line i of s code. i 2 ns , j ¼ f1; 2g, and 1 is descending, 2 is Upper line. During the coordinated period, the interval hsij between trains on the line Rsij is divided into two parts, and the departure time Lsij of each train on the line at the transfer station can be determined according to the departure time of the ﬁrst train on the line. The set of departure times Lsij can be expressed as ð1Þ

Lsij ¼ Lsij þ nsij hsij

nsij ¼ 0; 1; . . .; msij 1

ð2 1Þ

The ﬁrst train on the line Rsij should be satisﬁed ð1Þ

ta Lsij ta þ hsij The last train Rsij on the line should be satisﬁed

ð2 2Þ

330

Z. Xia et al. ð1Þ

Lsij þ ðmsij 1Þhsij tb

ð2 3Þ

According to the departure time of the train and the time of stay at the station, the arrival time of each train is set as n o ð1Þ Asij ¼ Lsij Dsij ¼ Lsij þ nsij hsij Dsij

ð2 4Þ

msij represents the number of trains running in the coordinated period. Formulas (22) and (2-3) indicate that the departure time of trains should be within the coordinated time period. Formulas (2-1) and (2-4) indicate the train’s basic operating parameters in the coordination period by the ﬁrst train. Therefore, the coordination of train connection at transfer station can be translated into the coordination of arrival and departure time of the ﬁrst train in each direction during the period. ðxÞ ðyÞ Taking transmission lines Rsij and connecting lines Rspq as research objects, On the line Rsij , x train passengers transfer to Rspq direction y train. To establish the connection, passengers must be allowed to get off the train and arrive at the connection line before the train has left. That is, the arrival time of the x train on the line Rsij plus the travel spq of the passengers arriving at the transfer line platform after getting off the train time tsij is less than the departure time of the y train in the direction Rspq . As ðxÞ

spq Asij þ tsij LðyÞ spq

ð2 5Þ

spq tsij refers to the passenger’s journey time from line Rsij to line Rspq . The value of spq is related to the setting of transfer mode, the length and width of transfer corridor, tsij the width of door and the flow of passengers. Usually, the transfer travel time of a certain direction can be taken as the average travel time of the passengers in that direction by investigation and statistics. In order to ensure that passengers leave the station by the ﬁrst train after arriving at the transfer platform, the time the transfer passengers arrive at the connecting line platform is longer than the departure time of the y 1 train in that direction. That is, ðxÞ

spq Lðy1Þ spq \Asij þ tsij

ð2 6Þ

According to formula (2-5) and (2-6), the transfer waiting time of the passengers in the transfer direction is ðyÞ

ðxÞ

spq ¼ LðyÞ wspq spq ðAsij þ tsij Þ sijðxÞ

ð2 7Þ

And the waiting time for passengers in the connection direction is. Therefore, the objective function can be expressed as

Transfer Optimization of Multi Standard Regional Rail Transit

minW ¼

X 8M

wspq sij ¼

mspq msij X X X Rsij !Rspq x¼1 y¼1

ðyÞ

ðyÞ

ðyÞ

wspq /spq cspq sijðxÞ sijðxÞ sijðxÞ

331

ð2 8Þ

ðyÞ

Where cspq denotes the number of passengers transferred from the x train on the sijðxÞ line Rsij to the y train on the line Rspq , and M represents the set of all transfer relations at the transfer station, and M 2 Rsij ! Rspq j i 6¼ p i; p 2 1; 2; . . .; ns j; q 2 1; 2 . 3.3

Algorithm Solution

From the analysis of the model, it can be seen that the optimization model of multi-system regional rail transit linking involves many parameters, and the solution of commercial software is more complex. In the optimization model solving algorithm, genetic algorithm is widely used in this ﬁeld, which can be designed to solve the model [7]. (1) Variable coding. The binary coding method is used to compile the binary chromosomes of all line trains in the network at the departure time of the transfer station. The coding length is determined by the range of values and the number of rail transit lines passing through the transfer station. (2) Generating initial population. In order to improve the quality of the initial population, the randomly generated individuals were screened to avoid the individuals not meeting the constraints into the initial population. The value of M is usually determined according to the size of the solution space. When the solution space is large, the value of M is also larger, otherwise the value is smaller. The general value range is 200–1000. (3) Determine ﬁtness function. Fitness function reflects the quality of individual targets. In this paper, the objective of the model is minimum type, and the ﬁtness function is designed as follows: FðxÞ ¼

Fm f ðxÞ Fm Fn

Among them, the objective function value is the maximum and minimum of the objective function. (4) Set up genetic operators. In this paper, the roulette selection method is used. The probability of a population being selected is pi ¼ Pfi f , non-sister chromatids i

i

(from both parents) often cross. In this paper, we choose the common single-point crossover method to randomly select a point in the parent’s chromosome and exchange the right genome of the chromosome with each other. In mutation operation, the mutation rate was 0.05. (5) Algorithm termination. The algorithm is terminated when the individual difference of the optimal ﬁtness function of the latter two generations is less than a certain range.

332

Z. Xia et al.

4 Case Analysis 4.1

Data Calculation

As shown in Fig. 2, a rail transit loop R1 in a city intersects a metropolitan railway line R2 with a transfer station e1 , which is calculated from 9:00–11:00 at peak period and 7:00–9:00 at morning peak.

Fig. 2. Schematic diagram of partial urban rail transit network connection

Peak hours 9:00–11:00, rail transit loop R1 direction of 300 s interval, outer ring direction 270 s; urban line R2 upstream direction 360 s, downstream direction 480 s (for the convenience of the latter 1 is used to indicate the inner ring or upward direction, 2 to indicate the outer ring or downstream direction). The travel time in all directions is 120 s, and passenger flow data are shown in Table 1. Table 1. Train changing passenger flow during peak hours R11 R12 R21 R22 R11 – – 100 210 R12 – – 110 200 R21 50 100 – – R22 100 50 – – R11 – The direction of the inner ring of rail transit.

In the case, the population size is 1000, the mutation probability is 0.9, and the crossover probability is 0.1. According to the algorithm designed before, it is solved by MATLAB 2016 programming. Find out the departure time of the two directions of the ð1Þ ﬁrst bus coordination period Lij in the station e1 . Table 2 is the departure time and waiting time of the 9:00–11:00 time train at the station where the model is solved.

Transfer Optimization of Multi Standard Regional Rail Transit

333

Table 2. The departure time of the ﬁrst bus during the peak period ð1Þ

ð1Þ

L11

L12

ð1Þ

L21

ð1Þ

L22

Waiting time(s)

Optimal solution 9:02:51 9:02:51 9:04:22 9:03:58 2.427 106 Initial solution 9:03:13 9:00:44 9:04:43 9:04:13 4.367 106 ð1Þ

L11 – The departure time of the ﬁrst train in the inner ring of rail transit.

Similarly, the peak time 7:00–9:00 is calculated. Interval h11 ¼ 180 s, h12 ¼ 180 s, h21 ¼ 240 s, h22 ¼ 270 s, transfer travel time is 120 s, stop time D11 ¼ D12 ¼ 30 s, D21 ¼ D22 ¼ 50 s, passenger flow data as shown in Table 3:

Table 3. Train changing passenger flow at rush hour R11 R12 R21 R22

R11 – – 480 600

R12 – – 480 650

R21 500 520 – –

R22 450 650 – –

The optimal solution and initial solution of the peak period are calculated as shown in Table 4.

Table 4. The departure time of the ﬁrst bus at the peak hour ð1Þ

L11

ð1Þ

L12

ð1Þ

L21

ð1Þ

L22

Waiting time(s)

Optimal solution 7:01:02 7:02:22 7:02:01 7:01:11 1.259 107 Initial solution 7:00:48 7:02:55 7:01:36 7:01:58 1.753 107

According to the stopping time and the distance between trains, the arrival time of each train trip in the coordination period can be determined. Based on the arrival and departure time of the metropolitan line, the operation line of the rail transit ring line is adjusted with the technique of the overall translation of the operation line to realize the optimization of the transfer and connection between the two lines. 4.2

Result Analysis

Tables 2 and 4 show that the optimal solution reduces the waiting time of passengers to a certain extent and veriﬁes the correctness of the model. The optimal coordination effect is expressed by the ratio of the total waiting time corresponding to the initial solution and the difference between the optimal solution and the initial solution. The relative optimization ratio of the peak period is 44% and the peak period is 28%. It

334

Z. Xia et al.

shows that the coordination and optimization effect of the peak period is good, which is consistent with the actual situation. During the peak hours, the passenger flow is small, the capacity is sufﬁcient, the interval between lines and workshops is large, the adjustable range is large, and the service level of passengers is enhanced. However, in the peak period, the passenger flow is large, the interval between lines and workshops is small, the waiting time for transfer is also small, and the optimization effect is not obvious.

5 Conclusion Multi standard rail transit is a major trend in the development of rail transit. From the point of view of transfer coordination, this paper optimizes the connecting time of various types of rail transit trains at transfer stations, which can reduce the waiting time of passengers, and veriﬁes its correctness with a case study. It has a certain reference value for the coordinated development of regional rail transit. However, this paper only considers the connection of various rail transit systems from a single transfer station, and the optimization of the connection of multiple transfer stations under the condition of network needs further study. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Zhou, Y.F., Zhou, L.S., Le, Y.X.: Study on synchronous coordination and optimization of train connection at urban rail transit stations. J. Railw. 33(3), 9–16 (2011) 2. Liang, Q.S., Li, X., Xu, R.H.: Optimization of train connection time at urban rail transit transfer station. Res. Urban Rail Transit 18(4), 9–13+41 (2015) 3. Tian, Y.F., Pan, H.C.: Train connection optimization method of urban rail transit transfer station based on service level. Urban Rail Transit Res. 20(7), 81–85+93 (2017) 4. Xu, R.H., Li, X.: Comprehensive optimization of the last bus connection scheme for urban rail transit network. J. Tongji Univ. 40(10), 1510–1516 (2012). (Natural Science Edition) 5. Xu, R.H., Zhang, M., Jiang, Z.B.: Study on the departure time domain of the ﬁrst and last trains of Urban Rail Transit Based on line network coordination. J. Railw. 02, 7–11 (2008) 6. Li, S.J., Xu, R.H., Yang, R.D.: Train operation plan optimization of urban rail transit network based on capacity coordination. J. Southeast Univ. 47(5), 1048–1054 (2017). (Natural Science Edition) 7. Wang, W.X., Chen, D.J., Chen, H.: Application of hybrid genetic algorithm in train flow allocation of railway network. Comput. Simul. 32(4), 129–132+153 (2015)

Research on the Time Differential Pricing of Intercity Railway Based on Congestion Pricing Theory Zhang Peng1,2,3, Zhou Xia1,2,3, Wang Minghui1,4(&), and Ni Shaoquan1,2,3 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, Sichuan, China [email protected], [email protected], [email protected], [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China 3 National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 Chongqing-Guizhou Railway Co., Ltd., Chongqing 400014, China

Abstract. This paper is inspired by the congestion pricing theory of urban road, which regulates the transportation demand to solve the imbalance between supply and demand at different times by the economic means. The paper utilizes the Beijing-Tianjin intercity railway as an illustrative example to analyze the feasibility of the model and algorithm. The result shows that time differential pricing are not only able to alleviate the contradiction between supply and demand in different periods, but also increase the income for the inter-city railway operation enterprises. Keywords: Intercity railway Time differential pricing

Congestion pricing theory

1 Introduction For a long time, the intercity railway in our country adopts the uniﬁed pricing method. However, this uniﬁed pricing method can not solve the contradiction of transportation supply and demand caused by the uneven distribution of passenger flow in time from the operation situation of Beijing-Tianjin intercity railway, Shanghai-Nanjing intercity railway and Guangzhou-Shenzhen intercity railway for many years. From the point of view of passengers, this contradiction is directly reflected in the fact that passenger travel experience is poor during the peak hours and even the demand for driving is not satisﬁed. From the perspective of intercity railway operators, this problem not only makes the transportation organization difﬁcult during the peak hours, but also directly affects the operation revenue. That is to say, passengers who require high comfort and timeliness may choose other modes of transportation because of poor ride experience, resulting in the loss of inter-city railway passenger flow during the peak period. However, in the case of excess capacity in non-peak hours, it is not effective in © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 335–342, 2019. https://doi.org/10.1007/978-3-030-04582-1_39

336

Z. Peng et al.

attracting passengers with low timeliness requirements from the peak period to transfer. Therefore, it is necessary to adopt a suitable pricing plan to achieve the “sharp peak ﬁlling” of passenger demand, and ﬁnally achieve a balance between supply and demand of transportation, and a win-win situation between inter-city railway operators and passenger groups.

2 Congestion Pricing Theory and Time Differential Pricing of Intercity Railway Congestion pricing theory is a theory of trafﬁc economy that has been put forward to solve the problem of urban road congestion since the mid 1970s. The theory of congestion pricing studies the contradiction between supply and demand imbalance by controlling and regulating the total amount of road demand through economic means under the existing scale of transportation supply, so that it can be redistributed in time and space [1]. Intercity railway is a passenger dedicated railway connecting adjacent cities and urban agglomerations. Usually, short distance passenger flow is the main train. The running time of trains is usually less than 1 h. On working days, the law of passenger flow between intercity railways and urban roads have certain similarities which is that its passenger flow consists mainly contain commuter passenger flow and public service passenger flow, as well as other passenger flows, such as entertainment, visiting relatives, personal affairs and so on. The commuter passenger flow has the obvious “tidal characteristics”, which is also the main reason for the formation of the peak in the morning and evening. This part of the passenger flow travel timeliness has certain requirements, but there are also some price-sensitive passengers to a certain extent willing to advance or wrong travel time to obtain a certain discount ticket. The demand of public service passenger flow also belongs to the rigid demand, the demand for travel immediacy is relatively strong, usually travel time in the two periods of 9:00– 11:00 and 14:00–16:00. This part of the passenger is not sensitive to the price, and the peak price increase during peak period will not affect their trip plan basically. Entertainment, personal affairs and other passenger trafﬁc demand for instant travel is not high. It may be scattered throughout the day. In the case of discounted tickets, a considerable number of passengers will be willing to accept the guidance of the discount policy and avoid peak trips. This part of the passenger flow will be the main audience of the time interval fare. The above analysis of the passenger flow law of inter-city railway shows that the passenger flow law of inter-city railway and urban road has certain similarities, and it has the conditions to use economic means to control the passenger flow of transport demand. At present, many scholars have applied the congestion pricing theory to the study of the peak time pricing of urban rail transit. At the same time, they put forward the peak price, the off-peak price reduction and the combination of the ticket pricing strategy [2]. However, the application of this method of dividing the operation period into peak period and off-peak period may bring new problems to the intercity railway: On the one hand, the deﬁnition of peak period and off-peak period is not as clear and

Research on the Time Differential Pricing of Intercity Railway

337

absolute as urban rail transit. On the other hand, the difference between the time of the peak period and the off-peak period is small and the price difference is more obvious than that of urban rail transit. Passengers may concentrate on traveling for a short period of time before or after the peak period, resulting in another peak in a short period of time before and after the peak period.

3 Model Establishment and Solution According to the relevant literature, railway fare development can be seen as a doublelevel programming model [4]. Railway operators are the upper decision makers, and passengers are the lower decision makers. 3.1

Restrictions

Transport Capacity Constraint. The transport capacity constraint means that the trafﬁc volume of each time period cannot exceed the sum of the maximum transport capacity of all trains during that time period. Assume that the maximum transport capacity per train is qmax , and the headway between the intercity railways is intended to be tk at each interval k, so nk ¼ 60=tk is the number of departures in time period k. And then passenger flow qk with limitation of transport capacity at each time interval can be expressed as 0 qk qmax nk ; k 2 T

ð1Þ

Fare Constraint. The Intercity railway, as a form of public transportation, its pricing is subject to restrictions and constraints by government. Related departments usually set a maximum fare pmax and a minimum fare pmin , so constraint for fares at every time interval is that pmin pk pmax ; k 2 T

3.2

ð2Þ

Model Design

Taking fare pk and passenger flow qk at time interval k as decision variables, maximizing railway proﬁtability as an upper target, and lowest travel expenses of travellers at each time interval as lower target, all of them synergistically achieve passenger flow balance in each period. Upper Model. Taking the fare pk as a decision variable, aiming at maximizing the proﬁt of inter-city railway operating companies, the upper level planning objective function is as follows:

338

Z. Peng et al.

maxI ðpk Þ ¼

X

qk ðpk Þðpk ck Þ; k 2 T

ð3Þ

k2T

In which Iðpk Þ is the proﬁt of the railway department with the fare pk , ck is the average transportation cost per passenger during the time interval. Lower Model. Taking the passenger flow qk as a decision variable, considering passenger travel expenses, and modeling the lower-level distribution model, it aims at the lowest travel expenses for passengers. Then the lower level planning objective function is as follows:

minUðqk Þ ¼

XZ k2T

qk

f ðqk Þd(qk Þ; k 2 T

ð4Þ

0

In the above formula, f ðqk Þ is the travel cost function of the traveler’s time periods. For intercity rail passengers, the difference in travel utility of travellers can be considered mainly due to the amount of fare paid and the waiting time at the station. So f ðqk Þ is composed with all pares Pk and sum of passenger waiting time Wk in time interval k. In other words, f ðqk Þ ¼ Pk þ Wk , and Pk ¼ pk qk . As for Wk ¼ Ck wk qk , Ck is the unit time value in time interval k, and its value depends on the degree of development of the local economy. wk is the expected time for the traveler to wait for the total time in time interval k. Assume that the passenger arrival time is random, and obeys the Poisson distribution of different strengths kk in each time period. According to the Poisson distribution theory, the expected value of passenger waiting time in each departure interval is Tk ¼ kk tk2 =2, and the total passenger waiting time expectation value in time period k is wk ¼ 60Tk =tk ¼ 30kk tk . Therefore, based on the above analysis, the ﬁnal expression of the lower target planning function is: minUðqk Þ ¼

XZ k2T

3.3

qk

30kk qk Ck tk þ qk pk dðqk Þ; k 2 T

ð5Þ

0

Solution Algorithm Design

The double-level programming problem is an NP-hard problem, and the solution is very complicated and there is no polynomial solving algorithm [4]. this paper intends to use the improved particle swarm optimization algorithm and the Frank-wolfe algorithm [5–7]. The main idea is to use the improved particle swarm optimization algorithm for the upper mathematical programming model, the Frank-wolfe algorithm for the lower mathematical model, and the iterative iteration through the upper and lower layers. The ﬁnal result is successively approximated to the optimal solution.

Research on the Time Differential Pricing of Intercity Railway

339

Similar to genetic algorithm, Particle swarm optimization is also based on a set of random solutions, and iteratively ﬁnds the optimal solution. Compared with the genetic algorithm, the particle swarm algorithm does not have too many parameters to be adjusted, and the algorithm process is simple and easy to implement, with high precision and fast convergence. Considering the premature convergence of particle swarm optimization, an interference factor is added to interfere with the optimal solution in the current group [6]. Assume q obeys the standard normal distribution q N ð0; 1Þ, and command optimal solution Pbest ¼ Pbest ð1 þ qÞ. And then the solution process can be prevented from falling into the local optimal solution and the global search ability of the algorithm can be improved. Therefore, the speciﬁc algorithm steps in this paper are as follows, where i represents the number of iterations, Gen represents the maximum number of iterations of the genetic algorithm, and X represents the i-th generation population. Step 1: Particle Swarm Algorithm Initialization (1) Initializing each parameter in the particle swarm algorithm; (2) The above model variables (fare discount rate or rate of increase for each time period) constitute the particles of the particle group, randomly initialize the T population size is n and the position is Xi ¼ x0i1 ; x0i2 ; ; x0ik and velocity of T each particle is Vi ¼ v0i1 ; v0i2 ; ; v0ik ; (3) Set the current position of each particle to be Lbest , and the position of the optimal particle in the population is recorded as Pbest . Step 2: Solve the underlying model Bring the solution Xij ¼ ði ¼ 1; 2; ; nÞ of the upper model into the lower model, and use Frank-wolfe to solve the optimal solution Yij ¼ ði ¼ 1; 2; ; nÞ of the lower model (the passenger flow in each period). Step 3: Calculate the ﬁtness function value Bring Xij ; Yij into the upper model, calculate the ﬁtness function value FðXij ; Yij Þ, that is intercity railway revenue I ðp; qÞ. Step 4: Update individual extreme value and group extreme value (1) If the ﬁtness function value corresponding to Xij is better than the ﬁtness function value of the current individual optimal position Lbest , then Lbest is updated to the new Xij , the corresponding lower layer’s optimal solution is updated to Yi ; (2) If the ﬁtness function value corresponding to Xi is better than the ﬁtness function value Pbest of the current global optimal position, then Pbest is updated to Xi , and the corresponding lower layer’s optimal solution is updated to Yij . Step 5: Determine whether the convergence condition is met. If yes, go to Step 8; if not, go to Step 6.

340

Z. Peng et al.

Step 6: Optimal solution interference. þ1 j ¼ Pbest ð1 þ qÞ, use the Frank-wolfe algorithm to solve the optimal Demand Pjbest jþ1 solution YPbest . Step 7: Perform speed update and position update on the particle swarm according to the following two formulas, and turn to Step 2. j vjikþ 1 ¼ xvikj þ a1 e1 ðlikj xikj Þ þ a2 e2 ðlgk xikj Þ

ð6Þ

xjikþ 1 ¼ xikj þ vjikþ 1

ð7Þ

Where a1 ; a2 2 R, generally, the value is 2, e1 ; e2 2 ½0; 1 is random number, likj is the j individual optimal particle position for the j-th iteration, lgk is the global optimal particle position for the j-th iteration, x is the inertia weight, and its value range is generally [0.1,0.9] [7]. Step 8: Output the optimal solution Pbest and YPbest of the upper and lower layer problems of the double-level programming, and output the upper and lower optimal function values to end the algorithm.

4 Case Analysis 4.1

Analysis of Fares and Returns

Take the Beijing-Tianjin intercity railway as an example to verify the feasibility of the model [5]. It’s obvious that in the case of a uniform fare, the passenger flow and the actual load rate of the passenger flow of the Beijing-Tianjin inter-city railway show obvious unevenness. So it ﬁts the conditions to implement a time-limited fare strategy. Calculated the ticket price and the passenger flow distribution under the condition of the ticket price through the double-level programming model and algorithm established above. The price reduction period of the time-limited fare corresponds to the period when the actual load rate is high under the original ﬁxed fare condition, and the fare reduction is just the opposite, which is in line with market rules. It Increases by 7.73% comparing with uniform pricing conditions, achieving an increase in operating companies. 4.2

Analysis of Actual Load Rate and Passenger Flow Situation

The model of Beijing-Tianjin inter-city train is CRH3C, and Fig. 1 shows the actual load rate at each time before and after the implementation of the split-time fare strategy. On the condition of ﬁxed fare, the actual load rate is maintained at 72.5%–112.0%. As for time-shared fare, the real load rate range is reduced to 84.2%–104.2%. It’s obvious that time-shared fares effectively utilize the transport capacity of existing operating trains, and to some extent, the congestion in the cabin is alleviated during peak hours.

Research on the Time Differential Pricing of Intercity Railway Unity constant result rate Minutes Time Statement Performance

120.0% Implementation rate

341

110.0% 100.0% 90.0% 80.0% 70.0% 60.0% 1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16

Time slot

Fig. 1. Actual load rate of each time period before and after the implementation of the timelimited fare

Figure 2 shows the passenger flow before and after the implementation of the timelimited fare, combined with the full load rate situation, we can ﬁnd that the implementation of time-shared fares plays a certain role in shaping the peaks and ﬁlling the valley, especially in the peak period of the two passenger flows, the effect of “shaving peak” is particularly obvious, which relieves the pressure of transportation organization during peak hours. In the 11th and 14th periods, the role of “ﬁlling the valley” is also more prominent, and the transportation capacity is more effectively utilized.

Constant customer service flow Partial timetable flow

Passenger flow / person times

4000 3500 3000 2500 2000 1500 1000 500 1

2

3

4

5

6

7

8 9 10 Time slot

11

12

13

14

15

16

Fig. 2. Passenger flow at each time before and after the implementation of the time-limited fare

342

Z. Peng et al.

5 Conclusion Taking Beijing-Tianjin inter-city railway as an example, compared with uniﬁed pricing, the calculation of Beijing-Tianjin inter-city railway under time-sharing pricing increased by 7.73%. The results show that the time-limited fare can not only play the role of “shaving the peak and ﬁlling the valley” to a certain extent, alleviating the contradiction between supply and demand of transportation in different time periods, but also increase the income for inter-city railway operators, which provides certain ideas for the rapid development of railway transportation pricing. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351,71761023), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017-RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Zhang, Z.H., Chen, L.R., Zhang, S.: Trafﬁc Investigation & Accounting, pp. 1–7. People’s Transport Publishing Company (2014) 2. Yang, W.J.: Basaltic system Numerical castles City orbit trafﬁc Temporary school charts Study summary research. Beijing Jiao-tong University (2016) 3. Zheng, P.J.: Under the condition of demanding demand Castle Road Temporal staircase constant research. Lanzhou Jiao-tong University (2015) 4. Chen, J.H., Gao, Z.Y.: The optimization strategy of railway ticket price based on double-level programming model is formulated, pp. 38–41. Bulletin of the Northern Jiao Tong University (2003) 5. Ben, A.O., Boyce, D.E., Lair, C.E.: A general bi-level linear programming formulation of the network design problem. Transp. Res. B 22(4), 311–318 (1988) 6. Zhao, Z.G., Gu, X.Y., Li, Y.T.: Comprehensive bilayer normal scale particle population priority algorithm. Practice of introducing systematic engineering principles, pp. 92–96 (2007) 7. Du, G.Z., Shi, Q.: Basics equilibriism principle stereotactic transport road safety concerns problem model solving algorithm. Syst. Phys. Rep., 469–474 (2009) 8. Lei, Q.Q.: Nonlinear nomenclature clever group superiority algorithm skill, pp. 7–10. Northern Ethnic University (2009)

Smart Vehicular Technology

Research of Fast Image Location Method Based on Improved Sobel Pingjun Zhang, Yang Liu(&), Yufeng Ji, and Xiaohong Wang School of Information Science and Engineering, FuJian University of Technology, Fuzhou 350118, China [email protected], [email protected], [email protected], [email protected]

Abstract. Current locating algorithm has achieved great research value in complex environment and different illumination, while the low velocity handicaps the application in real-time situation. In order to tackle the time consuming problem of image location, a fast algorithm which based on sobel edge check combined with FFT algorithm is put forward. Firstly, through applying the opencv method of image gray processing, image ﬁltering, sobel edge detection, binary image is extracted. Secondly, based on two respective space transformation: RGB to HSV, color-pair zone detection and edge location, this method can get the coarse position. Finally, for the purpose of getting accurate position, closure computing operation and geometry detecting is applied on the image. Through testing license plate, the algorithm has a great achievement in location accuracy and response performance. Keywords: Image location Shape detection

Fast sobel detection Color detection

1 Introduction Image location is an interdisciplinary discipline that integrates computer vision [1], multiview geometry, image retrieval [2], and machine learning [3]. It is also an important front-end link in the digital image recognition process, which mainly includes the collection, preprocessing, morphological operation, connectivity detection and accurate image positioning [3, 4] of digital image [5]. The current image location algorithm [6–8] has done a lot of research in all the above-mentioned links, highlighting improved methods such as gray-scale stretching, fast-directed ﬁltering and other methods have effectively improved the positioning accuracy. Edge detection [9] is the most studied by various researchers, the method based on eight directions of sobel edge detection locates well wherever the high-noise, but the speed of positioning is not high. These articles based on opencv [10–12] have a large number of library functions, which can guarantee the accuracy and stability of image positioning. The edge detection technology combining fast Fourier transform and sobel operator greatly improves the speed of image positioning.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 345–352, 2019. https://doi.org/10.1007/978-3-030-04582-1_40

346

P. Zhang et al.

According to the different nature of image processing, license plate location methods can be summarized as two classes based on grayscale and color images, where the positioning method of grayscale images [13] can be divided into edge detection, wavelet analysis, Genetic algorithm, mathematical morphology [14], license plate features etc. The key to the license plate location method based on color image [15] is the segmentation of color, but this method has great limitations because it is only applicable to lighting-uniform and background- simple. The research on license plate location based on HSV color space and mathematical morphology [14] proposed by Chang can quickly locate the license plate and the positioning accuracy rate is 96%, but it can’t accurately locate license plate photos with the vehicle color similar to the car; the opencv license plate image positioning [4] proposed by Hu, which adds illumination compensation, can accurately locate the license plate photos of the day and night and the positioning accuracy rate is 95%, however, this method can only locate the license plate photographs that meet the range of 2–4 m; A color-based image location algorithm [16] proposed by Wu achieves positioning under different lighting conditions and complex interference conditions, while the algorithm is complex and the positioning speed is slow; the license plate image location algorithm proposed by Sun based on color features and improved canny operator [13], combined with image color and edge features, can overcome the impact of light, positioning accuracy is 95% and the time is 0.052 s. This paper proposes an image positioning method based on opencv and fast sobel, through the test of 150 license plate images, the experiments results, whose accuracy rate was 98%, the average positioning time was 0.035 s, show that positioning performance was improved.

2 Edge Detection Based on Improved Sobel 2.1

Traditional Sobel

The sobel operator is mainly used for edge detection. Technically, it is a discrete difference operator used to calculate the approximate gray value of the image brightness function. In the process of algorithm implementation, the 3 * 3 template is used as a kernel to convolve and count each pixel in the image, then an appropriate threshold is selected to extract the edge. Using the 3 * 3 neighborhood avoids calculating gradients at interpolated points between pixels. The sobel operator is also a kind of gradient magnitude, i.e. the partial derivatives Gx and Gy can be implemented with a convolution template, as follows: 0

m Gx ¼ @ n m 0

m Gy ¼ @ 0 m

0 0 0 n 0 n

1 m nA m

ð1Þ

1 m 0 A m

ð2Þ

Research of Fast Image Location Method Based on Improved Sobel

347

The following formula is the result of horizontal convolution kernel and image convolution, where f is the pixel value of each point in the image. GX ¼ ðmÞ f ðx 1; y 1Þ þ 0 f ðx; y 1Þ þ m f ðx þ 1; y 1Þ þ ðnÞ f ðx 1; yÞ þ 0 f ðx; yÞ þ n f ðx þ 1; yÞ þ ðmÞ

ð3Þ

f ðx 1; y þ 1Þ þ 0 f ðx; y þ 1Þ þ m f ðx þ 1; y þ 1Þ GY ¼ ðmÞ f ðx 1; y 1Þ þ ðnÞ f ðx; y 1Þ þ ðmÞ f ðx þ 1; y 1Þ þ 0 f ðx 1; yÞ þ 0 f ðx; yÞ þ 0 f ðx þ 1; yÞ þ m ð4Þ f ðx 1; y þ 1Þ þ n*f ðx; y þ 1Þ þ m f ðx þ 1; y þ 1Þ The horizontal and vertical gray values of each pixel of the image are combined by the following formula to calculate the gray level of the point G¼

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ G2X þ G2Y

ð5Þ

The gradient direction can then be calculated using the following formula: h ¼ arctan

GY GX

ð6Þ

The sobel operator detects the edge based on the weighted difference between the upper and lower points of the pixel and the neighboring point, reaching the extreme at the edge. It has a smooth effect on noise, which provides more accurate edge direction information, while the edge positioning accuracy is not high enough. When the accuracy requirement is not very high, it is a more commonly used edge detection method. 2.2

Improved Sobel Edge Detection

Fast Fourier Transform Principle FFT [17] (Fast Fourier Transform) is a fast algorithm for DFT (Discrete Fourier Transform). The DFT calculation process as follows: XðkÞ ¼ DFT½xðnÞWNnk

k ¼ 0; 1; . . .; N=2

ð7Þ

DFT calculation analysis, a complex multiplication requires four real multiplications and two real additions, and a complex addition requires two real additions. The improved way of FFT is to make use of the symmetry, periodicity, reducibility, special points of the to make some items in the DFT operation merge able, which can decompose the long sequence DFT into short sequence DFT. Taking DIT- FFT algorithm (the Kuly-Tukey algorithm) as an example, the sequence x(n) is ﬁrst divided into the following two groups by the parity of N, then Eq. (1) can be decomposed into the following equations:

348

P. Zhang et al.

X ðK Þ ¼ X1 ðK Þ þ WNk X2 ðK Þ

ð8Þ

The DIT-FFT process greatly simpliﬁes the calculation of the DFT. The computational complexity of the DFT is N * N multiplications, N * (N − 1) additions (N represents the number of input sample points), while DIT-FFT The number of multiplications is N2 logN2 , the number of additions is N2 logN2 , the larger N, the more obvious the advantage of DIT-FFT algorithm is. DIT-FFT for Fast Sobel Convolution Let WN be the value of the horizontal or vertical direction operator, when WNx ¼ fm; 0; m; n; 0; n; m; 0; mg,WNy ¼ fm; n; m; 0; 0; 0; m; n; mg is a sequence of image pixel values, ZN is a sequence of image pixel values. ZN ¼

f ðx 1; y 1Þ; f ðx; y 1Þ; f ðx þ 1; y 1Þ; f ðx 1; yÞ; f ðx; yÞ f ðx þ 1; yÞ; f ðx 1; y þ 1Þ; f ðx; y þ 1Þ; f ðx þ 1; y þ 1Þ

ð9Þ

Since N = 9, not an integer power of 2, in order to be able to use the radix-2 algorithm, it is necessary to make 7 zeros so that N = 16. Using the radix-2 algorithm, for each image convolution operation, multiplication is performed 32 times and addition is performed 64 words. If a Fourier transform is used, multiplication is performed 81 times and addition is performed 72 times. The algorithm is tested on license plate image edge, the detection time of this algorithm is 0.020 s, while the traditional edge detection algorithm takes 0.024 s.

3 Image Positioning Algorithm Flow As shown in Fig. 1, the image processing flow in this paper mainly through image reading, grayscale, edge detection, vertical edge reﬁnement and other processes. The classic point of this paper is to vote for each pixel in the image to determine whether it belongs to the internal point of the license plate area, this step can extract the suspected area of the license plate; the determination of the license plate area requires the process of shape detection, morphological closed operation, and connected area detection. The voting mechanism described above, in particular, sequentially traverses the target image by using a 3 * 3 window. If there are at least two edge points and 8 blue pixel points in the range of 8 neighborhoods around the pixel, it can be judged that the pixel is an interior point of the license plate area, so that the interference in the nonvehicle area can be suppressed. The license plate ﬁne positioning described above performs morphological closing operation on the suspected points and connects the points set of each area. Then, the connected area detection is performed on the closed operation image, and the license plate area is screened in the outer contour of the connected area.

Research of Fast Image Location Method Based on Improved Sobel

349

Fig. 1. Image positioning algorithm flow

4 Results and Discussion Evaluation of the performance of this proposed algorithm. Tests were carried out on different sequence. The experimental environment of this algorithm is the windows system vs2012, opencv2.4.9, license plate positioning for 150 license plate photos, through multiple tests, 147 images can be accurately located, indicating that the algorithm is highly accurate. Using this algorithm to locate the license plate of a car, since the background color of the license plate is mostly blue, the shape of the license plate is rectangular, the aspect ratio is about 3, and the area of the character in the license plate accounts for about 20% of the entire area of the license plate, choosing the height threshold in the

350

P. Zhang et al.

HSV color space as (81, 38, 63), (135, 255, 255), the ratio of the aspect ratio of the license plate rectangle is between 2.5 and 3.5, and the non-zero pixel ratio inside the license plate area The proportion of the area to the entire area is set at 15% to 25%. Figures 2(a)–(h) shows the results of this algorithm. As shown in Fig. 2, 150 images of license plate are employed for test, all of them were taken by camera from different backgrounds in the real word. The results (Fig. 3) illustrates that 147 images can be located successfully, 3 images failed to position, most of the reasons for failing to locate the image correctly are because the scene is too complicated, the small part is caused by the body color being too similar to the license plate color (Fig. 4).

(a) Original image

(d) RGB to HSV space result

(b) Gray image

(c) Sobel edge detection

(e) Single-channel diagram (f) License plate candidate with blue threshold

(g) License Plate Area Determination

(h) License plate positioning

Fig. 2. The results of license plate location process

Research of Fast Image Location Method Based on Improved Sobel

351

Fig. 3. License plate images under different backgrounds

Fig. 4. A set of license plate positioning results

As shown in Table 1, comparisons between the accuracy and speed of positioning of different references and this algorithm can be seen, this paper combines DIT-FFT with sobel, greatly shortens the convolution time, and then accelerates the positioning rate. It is also determined in combination with morphological operations. The rectangular area of the license plate greatly improves the accuracy.

Table 1. Accuracy and times of location for licence plate by different algorithms Contradistinction Photo pixel (pixel * pixel) HM [3] 576 * 752 opencv [7] 576 * 752 canny [4] 576 * 752 DIT-FFT 576 * 752

Quantity Accuracy Time (s) 150 150 150 150

96% 95% 95% 98%

0.050 0.046 0.052 0.035

352

P. Zhang et al.

5 Conclusion In this paper, based on the improved fast sobel edge detection, an image positioning technique combining binarized image with image color and shape detection is realized. The license plate orientation is used as an experiment, which shows that the algorithm has higher accuracy and faster speed. Acknowledgement. The authors would like to thank the anonymous reviewers for their valuable ntcomments. This work was supported by Initial Scientiﬁc Research Fund of FJUT (GYZ11050), Fuzhou Science and Technology Fund (2018G84).

References 1. Zhang, J., Zhang, H., Liu, X.M., Zeng, S.Q.: A fast algorithm for locating the object point in binocular stereovision. Inf. Control 38(05), 563–570 (2009) 2. Yan, F., Zhou, C.J., Tian, Y.T.: Image edge point detection algorithm for object localization. J. Jilin Univ. (Eng. Technol. Ed.) 46(06), 203–211 (2016) 3. Peng, H.Q., Li, H.Y.: Robot visual orientation image processing. Comput. Technol. Autom. 21(01), 74–76 (2004) 4. Hu, X.H., Wang, B.T.: A open CV-based vehicle license plate location. Comput. Measur. Control 24(09), 206–210 (2016) 5. Li, P.: Research on the application and development digital signal processing technology based on the new situation. Electron. Test. 45(06), 76–77 (2016) 6. Sheng, X.Q., Guo, J.X., Zhi, Y.M., Huang, D., Zheng, X.L.: A fast-saliency method for realtime infrared small target detection. Infrared Phys. Technol. 35(03), 77–79 (2016) 7. Wang, L., Wen, D.S., Zhan, J.M.: A location method of transient light target. Energy Procedia 23(06), 13–16 (2011) 8. Sun, L.H., Zhao, E.L., Ma, L., Zheng, L.: An edge detection method based on improved sobel operator. Adv. Mater. Res. 32(09), 971–974 (2014) 9. Xiao, C.M., Shi, T.L., Xia, R.B., Wu, W.: Edge-detection algorithm based on visual saliency. Inf. Control 43(01), 9–13 (2014) 10. Chao, Y., Li, Z.J., Huang, S.F.: Programming and image processing based on the realization OpenCV. Electron. Des. Eng. 56(01), 74–76 (2004) 11. Qin, X.W., Wen, Z.F., Qiao, W.W.: Image processing based on OpenCV. Electron. Test. 43 (07), 39–41 (2011) 12. Li, Y.: Research on present status review of OpenCV. Ind. Control Comput. 30(07), 123– 126 (2017) 13. Sun, J.L., Pang, J., Zhang, Z.L.: Recognition of vehicle license plate locating based on color feature and improved canny operator. J. Jilin Univ. (Sci. Ed.) 53(04), 693–697 (2015) 14. Chang, Q.H., Gao, M.D.: Research on license plate location based on HSV color space and mathematical morphology. J. Graph. 34(04), 159–162 (2013) 15. Kim, K.B., Park, C.S., Woo, Y.W.: Recognition of car license plates using morphological features, color information and an enhanced FCM algorithm. Springer, Heidelberg, June 2007 16. Wu, W.X., Zhao, G.: Acolors based algorithm for license plate location. J. Inf. Secur. Res. 2 (01), 58–65 (2016) 17. Tian, X.H., Liu, W.J., Pei, X.M.: An algorithm for calculating the linear discrete convolution with FFT. Tech. Autom. Appl. 28(02), 56–58 (2009)

Study on Coordination Development Model of the Regional Rail Transit System Hao Huang1, Lan Liu1,2(&), Yi-Han Wang1, and Jian-Nan Mao1 1

2

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] National and Local Joint Engineering Laboratory of Integrated Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

Abstract. Studying the coordination development model of regional rail transit system and identifying the factors, which affects the coordination development of regional rail transit have signiﬁcant guiding efforts for further realizing the development of regional rail transit system. Based on the existing research results, the concept of the regional rail transit system is deﬁned. The Data Envelopment Analysis model is used to construct the regional rail transit system coordination development model. By taking Beijing as an example, the feasibility of the model was veriﬁed. The results show that the degree of coordination development of regional rail transit system in Beijing has been continuously improved during the sample period, but the coordination relationship is still not the most suitable, and the relevant suggestions are put forward. Keywords: Transport economy Coordination development

DEA model Regional rail transit system

1 Introduction With the rapid development of urbanization and rail transit systems, the rail transit system in a certain area has gradually changed from a single mode to a multi-mode coexistence mode. The rail transit operation mode has also gradually evolved from a single mode independent operation management mode. The rail transit operation mode is also gradually transformed from a single-mode independent operation management mode to a multi-mode integrated collaborative operation management mode. The coordination development of multi-modal rail transit integration has become the key to improve the overall capacity and comprehensive beneﬁts of regional rail transit. However, there are many common problems in different rail transit systems, such as inconsistent standards and poor cohesion. Therefore, it is of great signiﬁcance to identify the influencing factors of the coordination development of regional rail transit, and to explore the coordination development mechanism of regional rail transit system. The previous studies in the ﬁeld of coordination development of different systems mainly focus on the theory of coordination development and evaluating the model of coordination development. On the basis of system theory, Zhang discussed the © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 353–359, 2019. https://doi.org/10.1007/978-3-030-04582-1_41

354

H. Huang et al.

coordination problem of the trafﬁc transportation system, and made a preliminary study on the coordination development inter the sub-system and between the sub-system in [1]. And Zhang also discussed the basic steps of the coordination development and the coordination character of the transport system in [2]. Xue constructed a coordinative development evaluation model of the integrated transportation system and economic system, gave the criterion of coordination state, and quantitatively analyzed the degree of coordination between the two in [3]. Mu put forward an approach of evaluating synergetic development within a whole system and among subsystems by using DEA model in [4]. According to the approach proposed by Mu, Zhao discussed the synergetic development among the subsystems in railway, highway, and aviation in BeijingTianjin-Hebei in [5]. In summary, most of the current coordination development evaluation models are applied to composite systems. For the evaluation of the coordination development of transportation systems, most of the existing researches are coordination development evaluations of external systems such as transportation systems and operational environments or economic environments, there is little research on the evaluation of coordination development within the regional rail transit system. Therefore, this paper proposes a coordination development evaluation model for regional rail transit system based on DEA model, and the rest of this paper is organized as follows. In Sect. 2, the deﬁnition of the regional rail transit system is discussed. In Sect. 3, the DEA-based coordination development evaluation model will be introduced. A case study of Beijing regional rail transit system is presented in Sect. 4, which followed by conclusions in Sect. 5.

2 Regional Rail Transit System Deﬁnition In the past studies, there is not a clear deﬁnition of the regional rail transit system. Quan [6] proposed that the regional rail transit includes the national main railway, intercity railway, regional fast rail and urban rail transit, and put forward the method of dividing the regional track trafﬁc according to the difference of its function. Gao [7] deﬁned regional rail transit as the suburban railway that provided passenger transportation services within the city area, and considered it to be an integral part of urban rail transit. Zhen [8] deﬁned regional rail transit as intercity rail transit and suburban rail transit according to its operation features. In this paper, the regional rail transit system is deﬁned as an organic whole that provides passenger services consisting of rail transit employees, ﬁxed facilities, mobile equipment, and operations management, including general railways in the region (including existing lines, high-speed railways), suburban railways and urban rail transit. According to the difference of passenger transportation service provided by rail transit, since the city railway and urban rail transit mainly meet the passenger short distance transportation demand, it is considered that the city railway belongs to the part of urban rail transit, so the regional rail transit system can be divided into two parts, the railway subsystem, and the urban rail transit subsystem.

Study on Coordination Development Model

355

3 Coordination Development Model In a certain area, the railway subsystem provides service for long-distance travelers, while the urban rail transit subsystem provides service for passengers who travel short distances, railways and urban rail transit subsystem provide complementary passenger transport services. There are obvious synergy and complementarity relationship between the railway and urban rail transit subsystem. Therefore, the selection of the model must reflect this interdependent and mutually influential relationship. Since the DEA model can evaluate the degree of influence of internal or external factors of complex systems, the DEA model can be used to measure the relationship of the input and output of different subsystem. The DEA method takes each evaluation unit as a decision-making unit DMU, Let T DMUj ; j ¼ 1; 2; ; n be n decision-making unit, let Xj ¼ x1j ; x2j ; ; xmj 0 be the T input vector of DMUj , the output vector of each DMU is Yj ¼ y1j ; y2j ; ; yrj 0, m and r mean the number of inputs and outputs. The objective function of the typical DEA model is the efﬁciency index hj0 of DMUj0 , the DEA model can be described in model (1): 8 U T Yj0 > < Max hj0 ¼ V T Xj0 UT Y ð1Þ S:T: V T Xjj 1; j ¼ 1; 2; ; n > : U 0; V 0 For railway subsystem (Subsystem a)and urban rail transit subsystem (Subsystem b), the inputs and outputs index in model (1) are replaced by the inputs of Subsystem a and the outputs of Subsystem b, then the ‘cross’ inputs-outputs pairs that can be used to evaluate the degree of coordination development for Subsystem a and b is obtained. Each DMUj ; j ¼ 1; 2; ; n must meet the following characteristics: having the same objective and task, having the same external environment and having the same input and output index. Introduce a non-Archimedean inﬁnitesimal e into model (1), a linear programming model can be obtained as model (2): 8 min½Ch ða=bÞ eð^es þ eT s þ Þ ¼ VDe > > > n > P > > S:T: ka=bj xj þ s ¼ Ch ða=bÞx0 > > > j¼1 > > > n > P < ka=bj yj s þ ¼ y0 ð2Þ j¼1 > > n P > > > q ka=bj ¼ q; q ¼ 0 or 1 > > > j¼1 > > > ka=bj 0; j ¼ 1; ; n > > : s 0; s þ 0 In model (2), ka=bj is the unknown variable; s ; s þ are the relaxation variables; ^e ¼ ð1; 1; ; 1ÞT ; eT ¼ ð1; 1; ; 1ÞT . Equation (2) equals to a C 2 R model if q ¼ 0, otherwise, it equals to a C 2 GS2 model. T

356

H. Huang et al.

Let q ¼ 0 to solve the C2 R model ﬁrstly, the obtained Ch ða=bÞ is deﬁned as the coordination validity of Subsystem a to b that reflect the degree of coordination, and the development validity of Subsystem a to b is described as Eq. (3): X Dh ða=bÞ ¼ 1= ka=bj ð3Þ If Ch ða=bÞ ¼ 0, it indicates that the two subsystems are completely uncoordinated; If Ch ða=bÞ ¼ 1, it indicates that the input-output relationship of Subsystem a to b is the most suitable relatively; If Dh ða=bÞ\1, it indicates that increasing the input of the Subsystem a does not lead to a signiﬁcant increase in the output of the Subsystem b; If Dh ða=bÞ [ 1, it indicates that increasing the input of the Subsystem a can signiﬁcantly increase the output of the Subsystem b; If Dh ða=bÞ ¼ 1, it indicates that the output of the Subsystem b can keep a ﬁxed growth rate when increasing the input of the Subsystem a. If Ch ða=bÞ ¼ 1, it means that the C2 R model is efﬁcient, and then Dh ða=bÞ is equal to Ch ða=bÞ. Otherwise, let q ¼ 1 to solve the C 2 GS2 model to re-determine the coordination validity between the two subsystems of the corresponding DMU. We can obtain the coordination validity Ch ðb=aÞ and development validity Dh ðb=aÞ of Subsystem b to a as the same way. The coordination development validity of one subsystem to another has been calculated above, but they do not reflect the coordination and development validity between the two subsystems, so the coordination and development validity between Subsystem a and b are given as Eqs. (4) and (5): C ða; bÞ ¼

minfCh ða=bÞ; Ch ðb=aÞg maxfCh ða=bÞ; Ch ðb=aÞg

ð4Þ

Dða; bÞ ¼

minfDh ða=bÞ; Dh ðb=aÞg maxfDh ða=bÞ; Dh ðb=aÞg

ð5Þ

Let Sða; bÞ be the coordination development validity of Subsystem a and b, and the value of it is the multiplication of C ða; bÞ and Dða; bÞ[4], see Eq. (6): Sða; bÞ ¼ Cða; bÞ Dða; bÞ

ð6Þ

4 Results and Discussion 4.1

Data Preparation

According to the deﬁnition of regional rail transit system and the requirement of DEA model ‘cross’ inputs-outputs pairs, considering the scientiﬁcity, validity, rationality and availability of the data, we choose the ﬁxed investment, total length of the lane, the numbers of employees, the number of trains and the average speed of train travel as the

Study on Coordination Development Model

357

inputs of railway and urban rail transit subsystem, while the output indicator is passenger volume. Taking Beijing regional rail transit system as an example, choosing the related data of Beijing railway and urban rail transit subsystem from 2003 to 2015, this paper evaluates its coordination development level. 4.2

Results and Discussion

This paper takes 13 years from 2003 to 2015 as the decision-making unit, uses Matlab software to solve the DEA linear programming model, and obtains the coordination development validity of Beijing railway and urban rail subsystem, see in Table 1 and Fig. 1. Table 1. Coordination development validity between Beijing railway and urban rail subsystem Year Sða; bÞ Year Sða; bÞ

2003 0.189 2010 0.530

2004 0.189 2011 0.647

2005 0.207 2012 0.770

2006 0.208 2013 0.994

2007 0.194 2014 1

2008 2009 0.353 0.411 2015 0.981

Fig. 1. Coordination development validity between Beijing railway and urban rail subsystem

As can be seen from Table 1 and Fig. 1, there is a good coordination development relationship between Beijing railway and urban transit subsystem. According to the change rule of coordination development validity, these 13 years can be divided into 3 stage: low coordination development stage, rapid-development stage and high coordination development stage. In the low coordination development stage, the passenger transportation of Beijing railway subsystem is still dominated by the existing lines, while the urban rail transit subsystem is in the initial stage of construction, with a small number of lines and short operation mileage, and the backward infrastructure is the

358

H. Huang et al.

reason for restricting the coordination development of these two subsystems. In the rapid-development stage, on the basis of the rapid development of infrastructure, the improvement of operation and organization level is the essence of the rapid improvement of these two subsystem, which is intuitively reflected in the increase in the number and the speed of trains, the reduction of headways, and the improvement of other technical factors. In the high coordination development stage, these two subsystems basically realized the complete connection of the infrastructure, and has reached a high level of coordinated operation and organization, but the coordination development relationship is not completely suitable from the view of DEA model. Therefore, other means can be put into effect from policy, regulations, ticket service, security checking, resource information and other aspects to improve the coordination development level of Beijing railway and urban rail transit subsystem.

5 Conclusions In this paper, a DEA model is proposed to evaluate the coordination development degree of regional rail transit system. Taking Beijing as an example, the results show that The coordination development validity of the sample period can be divided into 3 stages: low coordination development stage, rapid-development stage and high coordination development stage. The backward infrastructure is the reason for restricting the coordination development of Beijing railway and urban rail transit subsystem in the low coordination development stage; the improvement of operation and organization level is the essence of the rapid improvement of these two subsystems in the rapiddevelopment stage; the degree of coordination development could be improved from the perspectives of policy, regulations, ticket service, security checking, and resource information in the high coordination development stage. Acknowledgement. This work is supported by National Key R&D Program of China (2017YFB1200702).

References 1. Zhang, S.R., Yan, B.: Analysis of the transportation system coordination. J. Chang’an Univ. (Nat. Sci. Ed.) 22(2), 51–53 (2002) 2. Zhang, S.R., Shao, C.F.: Coordination theory and model of transportation system. Math. Pract. Theor. 37(6), 1–6 (2007) 3. Xue, F., Zou, B.: Coordination analysis between transportation systems and economic systems. J. Transp. Eng. Inf. 15(04), 61–66+86 (2017) 4. Dong, M.U., Zhi-Ping, D.U.: A study on DEA evaluation for synergetic development of system. Math. Pract. Theor. (04), 56–64 (2005) 5. Zhao, L., Liu, J.: DEA evaluation study about synergetic development of regional transportationin Beijing-Tianjin-Hebei. J. Beijing Jiaotong Univ. 40(1), 124–129 (2016)

Study on Coordination Development Model

359

6. Quan, R.S., Liu, J.F.: Issues and thoughts on regional rail transit planning. Urban Transp. China 15(01), 12–19 (2017) 7. Gao, F., Lei, L., Yu, P.: Analysis on the joining mode of urban and regional rail transit. Railw. Transp. Econ. 31(08), 56–58 (2009) 8. Zhen, S.Q.: Construction standards and operational management of regional rail transit. Urban Mass Transit 19(S1), 29–31+35 (2016)

Research on Comprehensive Evaluation Method of Regional Railway Network Scale Feng Zhao1, Hong-Xia Lv1,2,3(&), and Bing Wang1,2,3 1

3

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest JiaoTong University, Chengdu 610031, China National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest JiaoTong University, Chengdu 610031, China

Abstract. In 2016, the “medium- and long-term railway network planning” was considered and adopted at the 139th executive meeting, marking the impending completion of the grand blueprint for the development of China’s railway network. In the process of improving and constructing the railway network, the comprehensive evaluation of regional railway network scale becomes particularly important. This paper analyzes the factors affecting the scale of regional railway network, determines the indicators of the comprehensive evaluation system of regional railway network scale, and analyzes it by Grey Relational Analysis. Finally, we take the Jing-Jin-Ji area as an example and give some opinions and suggestions. Keywords: Regional railway network scale Grey relational analysis

Comprehensive evaluation

1 Introduction The scale of the railway network is an important part of the railway network construction and regional layout planning. The development of regional railways can reduce the pressure on road trafﬁc in urban areas, reduce pollution and Dependence on automobiles and oil, and also drive the development of the area into a residential area of the city. The common problem in the transportation industry and the world social and economic development is regional imbalance development. So, the comprehensive evaluation of the reasonable regional railway network scale will contribute to the construction of the railway network in the long-term region, the improvement of the industrial layout and the development of the regional economy. The comprehensive evaluation of the regional railway network scale can effectively achieve regional and regional convergence, improve the integrity of the railway network, improve the operational efﬁciency of the road network, and promote the further development of regional and regional social and economic development.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 360–368, 2019. https://doi.org/10.1007/978-3-030-04582-1_42

Research on Comprehensive Evaluation Method

361

At present, most scholars study from a certain angle in the regional railway network. Therefore, comprehensive evaluation and analysis of the regional railway network will help improve the scientiﬁc and reliability of the railway network planning. Ou Guoli, Cui Denghui argued that the scale of the small railway network could not meet the increasing transportation demand of the whole society, and pointed out that the determination of the reasonable railway network scale was of great signiﬁcance to economic and social development [1]. Liang Dong used its innovative method to measure the scale of the railway network. It was estimated that the economic scale of the reasonable railway network would be 180,000–200,000 km in 2030 [2]. Based on the node importance and gravity model, Dong Weihua put forward corresponding suggestions for the planning and construction of the railway network in the metropolitan area [3]. Yang Jingshuai calculated the reasonable scale of the network from the perspective of urban trafﬁc demand and network service level [4]. Summarizing the existing literature, the railway network scale comprehensive evaluation system with multiple impact indicators is a typical gray system. The evaluation process has method diversity and uncertainty of results, so it is appropriate to use grey theory for evaluation. This paper determined the indicators of the comprehensive evaluation system, and used the data of recent years to analyze the grey relational analysis of each factor of the regional railway network system, and gave relevant suggestions on the long-term development of the regional road network based on the results.

2 Establishment of Comprehensive Evaluation Index for Regional Railway Network The indicators of the comprehensive evaluation system can be represented by some representative relative numbers, averages, or absolute numbers. The indicators within the evaluation system need to reflect the different influencing factors of the regional railway network scale. And according to the research purpose and the object, a series of selected representative indicators are combined to form an indicator system. The regional railway network comprehensive evaluation system indicators are as follows: (1) Per capita GDP GDP is an abbreviation for per capita GDP, which is a very important indicator used in development economics to measure economic conditions. This indicator can clearly show the economic operation of a region. The economic development of a region directly affects whether the construction of the transportation infrastructure in the region is perfect or not. Therefore, the scale of the regional railway network is closely related to the per capita GDP. Calculation formula: Per capita GDP ¼

Total output value total population

362

F. Zhao et al.

(2) The total retail sales of social consumer goods This is a retail sales of consumer goods that directly trade with urban and rural residents or units in the lodging, wholesale, catering, and retail industries and other industries. It is the cost that the urban and rural people consume in their lives, as well as the sum of the costs consumed by schools, hospitals, enterprises and troops in this area. This indicator mainly reflects the demand for consumption within the region. Through the change of this data, the regional railway network can be reasonably and appropriately adjusted to meet the needs of regional people. (3) End of year balance of urban and rural residents’ savings deposits, Residents’ Consumption Level. End of year balance of urban and rural residents’ savings deposits refers to the balance of individuals in urban and rural areas at the end of the year. Residents’ Consumption Level refers to the extent to which residents consume in the process of satisfying their own lives and enjoyment. The above two data can clearly reflect the consumption demand and consumption ability of the residents in the region. Further analysis and research can make the regional railway network more reasonable Calculation formula: Gross domestic consumption in the gross domestic product during the reporting period Residents' Consumption Level ¼ Annual average population of the report

(4) Total ﬁxed assets investment This indicator represents the amount of activity in the acquisition and construction of ﬁxed activity assets. And generally come out through money. It is a comprehensive indicator reflecting the relationship between the scale, speed and proportion of investment in ﬁxed assets. This indicator reflects the largest investment quota that can be utilized in the construction of a regional railway network. According to this data, the scale of regional railway network can be reasonably constructed, expanded and improved. (5) Railway passenger freight trafﬁc volume Railway passenger volume is the total number of passengers sent or arrived by the Railway Bureau or Railway General Corporation within a certain period of time. Railway freight volume is the total number of passengers sent or arrived by the Railway Bureau or Railway General Corporation within a certain period of time. These two indicators are important factors in the choice of route plan, the basis for evaluating the economic beneﬁts of railways, and the basis for designing railway capacity. Through this data, we can get a good understanding of the railway department operation in this region during a certain period of time.

Research on Comprehensive Evaluation Method

363

(6) Railway passenger and freight turnover The railway passenger turnover is the product of the number of passengers transported by the railway and the distance traveled within a certain period of time. The railway freight turnover is the product of the quantity of goods in the railway transportation and the distance of transportation within a certain period of time. This data can clearly show the situation of the railway department in the process of passenger and freight delivery. Calculation formula: Passenger turnover ¼ Actual number of passengers transported average journey per passenger Freight turnover ¼ Tonnage of actual shipment average haul of freight traffic

(7) Year-end population in each region This indicator refers to the population of the area before 24 o’clock on December 31 of each year. It usually reflects the population of this area intuitively. This indicator indirectly reflects the level of urbanization in this region, which can be combined with this data to increase or decrease the services provided by the railway to maximize the transportation needs of all people. (8) Urban per capita disposable income This indicator is the amount of money that everyone in the town can use for their own control. It helps the railway department to rationally adjust the price charged by the transport and transportation services, so that passengers and freight owners can improve their satisfaction. This will make the transportation services provided by the railway more popular. Gradually increase the scale of the regional railway network. (9) The ﬁrst, second and third industry output value The ﬁrst industry is: agriculture, forestry, animal husbandry and ﬁshery. The second industry is mining, gas, heat, manufacturing, electricity, construction, supply industry and water production. The third industry refers to other industries besides the ﬁrst and second industries. The output value of the three industries indicates the gross domestic product of the region. And can be compared with each other to know the development trend of this area. Through this data, the corresponding output value can be reasonably improved to further improve the scale of the railway network. According to the availability and representativeness of each data in this study, the indicator system of the factors affecting the scale of the regional railway network is: There are 4 ﬁrst-level indicators: Social and economic development level, railway transportation demand indicators, population density and distribution indicators. Divide each of the ﬁrst-level indicators into several secondary indicators, Fig. 1.

364

F. Zhao et al.

Regional Railway Network Scale Influencing Factors Index System

Target

Social and economic development level

First class title

railway transportaƟo n demand indicators

Railway freight turnover

Railway passenger turnover

Railway passenger volume

Railway freight volume

Urban per capita disposable income

Primary industry output value

Rural per capita net income

Year-end population in each region

Total fixed assets investment

Residents ‘ConsumpƟon Level

End of year balance of urban and rural residents' savings deposits

The total retail sales of social consumer goods

Per capita GDP

Secondary title

populaƟon density and distribuƟon indicators

Fig. 1. Index system for the factors affecting the scale of regional railway network

3 Comprehensive Evaluation Method for Regional Railway Network Scale Grey relational analysis (GRA) is a method to quantitatively describe and compare the development trend of the system. The basic idea is to determine whether the connection is tight by deﬁning the reference data column and the degree of geometric similarity of several comparison data columns, which reflects the degree of association between the curves. This method does not need to consider the size of the sample, and the same applies regardless of whether the sample is irregular or not. Generally, it is difﬁcult to determine the number of factors affecting the regional railway network scale and the law between those factors. Therefore, it is suitable to use the grey correlation analysis method for comprehensive analysis and evaluation [5]. The steps to analyze the gray correlation analysis method are as follows: (1) Identify reference sequences and comparison sequences A sequence of data that reflects the behavioral characteristics of the system is referred to as a reference sequence. Other data sequences used for computational analysis are referred to as comparison sequences. (2) Uniﬁed the dimension of the reference sequences and comparison sequences Due to the lack of comparability between different dimensional data, the raw data needs to be dimensionless. (3) Find the grey relational coefﬁcient n(Xi) of the reference series and the comparison series

Research on Comprehensive Evaluation Method

365

The so-called degree of association is essentially the difference in geometry between curves. Therefore, the difference between the curves can be used as a measure of the degree of association. For a reference sequence X0, there are several comparison sequences X1, X2,…, Xn, and the relational coefﬁcient n(Xi) of each comparison sequence and the reference sequence at each moment (i.e., each point in the curve) can be calculated by the following formula: n0i ¼

DðminÞ þ qDðmaxÞ D0i ðkÞ þ qDðmaxÞ

Where q is the resolution coefﬁcient, generally between 0 and 1, usually taking 0.5. D(min) is the minimum difference of the second level, D(max) Is the maximum difference between the two levels. Doi(k) is the absolute difference between each point Xi on the curve of comparison series and each point X0 on the curve of the reference sequence. (4) Find the relational degree Because the correlation coefﬁcient is the correlation value between the comparison sequence and the reference sequence at each moment (i.e., each point in the curve), it has more than one number, so the information is too scattered to facilitate the overall comparison. Therefore, it is necessary to concentrate the correlation coefﬁcients at various moments (i.e., points in the curve) into a single value, that is, use the average value as the quantity representation of the relational degree between the comparison sequence and the reference sequence. The formula for the relational degree ri is as follows: ri ¼

N 1X n ðkÞ N k¼1 i

ri represents the gray relational degree between the comparison sequence Xi and the reference sequence X0, which is also called the line relevance degree, the sequence relevance degree, and the average relevance degree. The closer the value of ri is to 1, the better the correlation is. On the other hand, the worse the correlation is. (5) Relevance ranking The degree of association between factors is mainly described by the order of relevance, not just the numerical value of the relational degree. First, sort the relevance of m subsequences to the same parent sequence in order of magnitude, denoted as {x}, The relevance degree reflects the “good or bad” relationship between subsequences for the parent sequence. If r0i > r0j, {xi} is bette r than {xj} for the same sequence {x0}, denoted as {xi} > {xj}; r0i denotes the eigenvalue of the i-th subsequence to the previous sequence.

366

F. Zhao et al.

4 Case Analysis Beijing-Tianjin-Hebei is China’s “Capital Economic Circle”. After continuous development, Jing-Jin-Ji area has gradually optimized the spatial layout of the city, improved the urban ecological environment, innovated the city’s management system, improved the city’s governance system, and constantly improved the city’s environmental quality and people’s quality of life [6]. Through the gray correlation analysis of the railway mileage, The third industry added value, the per capita GDP, the freight turnover, the passenger turnover, the passenger volume and the freight volume in the Jing-Jin-Ji area, according to the results obtained, relevant opinions and suggestions can be put forward for the long-term planning of the road network. 4.1

Calculation Results

The relational coefﬁcient between the various indicators of the regional railway network comprehensive evaluation system and the scale of the railway network is shown in Table 1. The third industry added value, per capita GDP, passenger turnover, freight turnover, passenger volume, and freight volume on railway mileage are 0.63, 0.70, 0.89, 0.83, 0.91, and 0.96. Table 1. Relational coefﬁcient Years

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

The third industry added value 1.00 0.97 0.92 0.86 0.80 0.74 0.66 0.60 0.55 0.50 0.44 0.41 0.38 0.36 0.33

Per capita GDP 1.00 0.98 0.93 0.87 0.82 0.78 0.71 0.66 0.64 0.59 0.54 0.52 0.50 0.49 0.48

Passenger turnover

Freight turnover

Passenger volume

Freight volume

1.00 1.00 0.96 0.92 0.88 0.85 0.80 0.79 0.80 0.75 0.71 0.73 0.72 0.73 0.83

1.00 0.98 0.95 0.98 0.97 0.95 0.92 0.90 0.88 0.85 0.82 0.82 0.81 0.77 0.80

1.00 0.98 0.93 0.99 1.00 0.99 0.97 0.93 0.90 0.87 0.85 0.85 0.82 0.78 0.80

1.00 0.99 1.00 0.98 0.95 0.98 0.94 0.93 0.94 0.96 0.94 0.95 0.96 0.98 0.92

Research on Comprehensive Evaluation Method

4.2

367

Relevant Recommendations Based on Calculation Results

Through the above calculations, the most important factor affecting the development of the Jing-Jin-Ji area is the volume of freight. The scale of the Jing-Jin-Ji regional railway network increased signiﬁcantly from 2001 to 2015. In 2015, the railway mileage increased by 56.11% compared with 2001, and also increased by 8.3% compared with the previous year. The scale of the regional railway network has increased signiﬁcantly, which has also led to The increase in the density of railway networks. The continuous improvement of the railway network is also promoting the development of the regional economy. In 2015, the passenger turnover of the regional railway network increased by 127.44% compared with 2001, and the freight turnover increased by 117.1% compared with 1996. These improvements indicate a signiﬁcant increase in the capacity of rail network transportation. In addition, from 2001 to 2015, the per capita GDP also increased signiﬁcantly, while the output value of the tertiary industry also showed an increasing trend year by year. Therefore, for the Jing-Jin-Ji area, it is necessary to strengthen organizational leadership and road network planning and design, promote the deep integration of cross-regional civil aviation, railways, highways, ports and other means of transportation, and improve the constitution of comprehensive transportation development in the Jing-Jin-Ji area. Therefore, for the Jing-Jin-Ji area, it is necessary to strengthen organizational management, plan and design the road network reasonably, promote the deep integration of cross-regional civil aviation, railways, highways, ports and other means of transportation, and improve the constitution of comprehensive transportation development in the Jing-Jin-Ji area. In addition, as the key point of the coordinated transport function of Beijing, railway transportation should be based on the convenience of industrial function adjustment and residents trip, at the same time, it is necessary to improve the rationality and scientiﬁcity of the rail transit project and related railway construction. In the process of comprehensive development of transportation, it is necessary to strengthen cooperation between various modes of transportation, break the existing institutional obstacles, and aim at comprehensive development, and promote the integrated development of inter-city railways, operating agencies, and national trunk lines. Finally, we must unify service and standard speciﬁcations, promote the standardization of transportation infrastructure, improve the level of regional transportation organization and infrastructure utilization, and apply advanced technologies such as the Internet and big data to the construction of railway networks to comprehensively enhance railway network construction of Jing-Jin-Ji area.

5 Conclusions This paper had comprehensively evaluated the scale of the regional railway network from the overall level. The research content was mainly divided into three parts: The ﬁrst part studied the constraints of the regional railway network and integrates and sorts them out. The second part analyzed the obtained indicators to determine the comprehensive evaluation method. In the third part, the gray relational analysis of the

368

F. Zhao et al.

influencing factors of the regional railway network was carried out, and the factors most related to the development of the road network scale were analyzed as the main factors. Through the calculation of the relevant data of the railway network in Jing-JinJi area, the corresponding opinions and suggestions are given, and the feasibility of this method is also proved. Due to the limited conditions, the evaluation methods need to be improved. Future research can consider in-depth analysis from this aspect. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project NO. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Ou, G.L., Cui, D.L.: Analysis of China railway network density and transportation load. China Transp. Rev. 10, 13–16 (2005) 2. Liang, D.: Thoughts on the scale of railway network. Railw. Econ. Res. (2017) 3. Dong, W.H.: Research on the evaluation of the reasonable scale and layout for metropolitan railway network—with the Jing-Jin-Ji Metropolitan as an Example. Beijing Jiaotong University (2010) 4. Yang, J.S.: Rational scale & layout methods for urban rail transit network. Southwest Jiaotong University (2006) 5. Kuo, Y., Yang, T., Huang, G.W.: The use of grey relational analysis in solving multiple attribute decision-making problems. Comput. Ind. Eng. 55(1), 80–93 (2008) 6. Zhou, L.Q., Ding, K.H.: Binhai new area and the rising of Jingjinji metropolitan area. J. Tianjin Norm. Univ. (2007)

Research on the Development Status of the Heavy-Duty Truck in China and Improving Its Safety and Energy Consumption Feifei Wei1(&), Hongye Pan2, and lv Hongxia1 1

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China

Abstract. With the generous use of trucks, fuel consumption and security risks increase. In this paper, we developed an Omni-directional steering electric truck model to improve the driving safety and reduce fuel consumption. The electric truck model is driven by four in-wheel motors which run independently. Then the dynamic model of the traditional truck is analyzed. Combining road environment information and the truck driving condition, the adaptive cruise control (ACC) system controls the truck drives at a safe speed. Using the ACC technology in electric trucks results in the improvement of energy efﬁciency and safety of trucks at the same time. Keywords: Heavy-duty truck

Omni-directional steering ACC

1 Introduction With the development of highway and rapid economic growth, China’s heavy-duty truck transportation developed by leaps and bounds. And transportation links between cities have been formed. There are about 75% of the goods transported by trucks in China, also the proportion is up to 90% in the United States and Europe [1]. As one of the most important tools of logistics, trucks sensitively reflect the changes of the economy, also truck industry development plays an important role in the development of the social economy, especially for the China’s logistics industry. Nowadays, heavy-duty trucks industry is facing the threat of market economic structure adjustment, vehicle price reduction and industry average proﬁt decline [2]. Since the impact of the ﬁnancial crisis in 2008 and the exiting of the automobile stimulus policy in 2010, there is a substantial decline in truck sales [3]. And with the decline of global oil reserves as well as the impact of environmental pollution control, the truck market will cope with a major challenge. So it is needed to improve the market competitiveness by improving the safety and energy efﬁciency of the heavy-duty truck. The development of the truck market is mainly influenced by the two factors of the trade policy and the economic environment [4]. © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 369–377, 2019. https://doi.org/10.1007/978-3-030-04582-1_43

370

F. Wei et al.

With the popularity of heavy-duty trucks, it brings many grave problems of energy and environmental protection, such as consuming too much fuel, emitting signiﬁcant pollutant and greenhouse gas, which are waiting to be solved. According to statistics, 4.35 million heavy-duty trucks consume more fuel than 83 million cars [5]. Therefore, the government has carried out policies to restrict truck fuel consumption [6]. Because of the particularity of the use of large trucks, it determines the obvious particularity of trucks in volume, weight, structure, and many other aspects compared with general vehicles. These cases increase the difﬁculty of driving, as well as security risks. About 80% trafﬁc accidents are due to the driver’s miscarriage of justice and improper operation [7]. The main accident types are collision with ﬁxture and trailing collision which respectively account for 45.10% and 39.22% of the total number of accidents [8], common reasons of such accidents including over-speeding, overloading, overtaking by drivers and the lateral instability of the vehicle, etc. It can be summed up China truck market has the following characteristics: (1) the political and economic environment continues to influence the heavy-duty truck market; (2) freight price is falling, yet oil price is rising; (3) safety, intelligent demand exuberant. Therefore, future heavy-duty truck market competition is not only the cost scale competition, product upgrading, technological progress, new energy vehicle, intelligence and modern logistics and networking will be the focus of competition.

2 System Structure 2.1

In-Wheel Motor Driving Electric Truck

In recent research, in-wheel motor is one of the hot spots of advanced electric vehicle technology [9]. The in-wheel motor driving an electric vehicle can be realized independent steering controlling, such as the vehicle steering operation performance is greatly improved [10]. Comparing the performance of various motors, the brushless DC motor is chosen as the driving motor. Referred to the traditional driving mode and steering mechanism, four in-wheel motors is adopted. The chassis mechanical model of the in-wheel motor driving electric truck is designed and shown on Fig. 1. This electric truck consists of four in-wheel driving motors and four sets of independent steering mechanism. By combining independent wire control steering technology and traditional mechanical steering mechanism, it formed a new type of Omnidirectional steering system, which improves the operational flexibility and stability. Figure 1 shows the design diagram of the in-wheel motor driving electric truck. In this electric truck system, 4 in-wheel motors are used as the driving wheel. The parameters of the driving motor are calculated by analyzing the dynamic model of the traditional truck, and the Brushless DC motor is chosen as the driving motor. The control model is established based on the motor mathematical mode, and DSP TMS320F2812 controller is used as the control platform [11] to analyze and evaluate the results. According to the environment information of the truck driving process, the adaptive cruise control model is established to control the speed to improve the safety and intelligence. Meanwhile, the electric vehicle has a new type of Omni-directional steering system. In the normal traveling, the steering wheel controls the front two sets

Research on the Development Status of the Heavy-Duty Truck

371

of steering mechanism to achieve the steering. When traveling in the crowded urban road or in narrow parking space, the driver can command the shift button to control the action of four sets of Omni-directional steering mechanism. The electric vehicle can achieve Omni-directional motions such as lateral parking (LP) and zero radius turning (ZRT). Thus it can improve the steering flexibility of the vehicle.

Fig. 1. Diagram of the whole system of the electric truck.

Through coordinating the four in-wheel driving motors’ speed and torque, the vehicle can be driven normally running and achieved the cruise control of the speed. Four steering motors can achieve Omni-directional motions to make the operation more stable in the steering mode. 2.2

ACC System of Electric Truck

Automation and intelligentization have become the development trend of the vehicle industry. The vehicle adaptive cruise control (ACC) system is one of the important advanced driver assistance systems (ADAS) to improve the ride comfort and safety [12]. Along with the EVs developed, ACC is essential to conﬁgure on Evs which is planned as a driver-assistance system controlling the speed automatically. The user vehicle always keeps with the same speed of the front one. When the vehicle encounters the red light or parking congestion, the system can also control the vehicle to stop quickly and start as the front vehicle moving. The system structure is shown in Fig. 2.

372

F. Wei et al.

Fig. 2. Adaptive cruise control system framework.

ACC system contains three sub modules. The information exchange module exchanges information of the vehicle and the ones nearby, such as weather condition, vehicle load and trafﬁc information. The control module includes the electronic control unit (ECU), motor control module, motor driver, battery pack, energy management system and brushless DC motor. ECU issues instructions to the motor control module and energy management system to achieve the vehicle control response. When the vehicle is deceleration or braking, the motor changes into motor generator to charge the battery pack to achieve energy recovery. The operation instruction module acts as a bridge to carry on the information transmission of the driver and the car.

3 Theoretical Analysis The safety spacing distance refers to the minimum distance between the traveling vehicles to ensure the trafﬁc safety. And there is no collision trafﬁc accident if two contiguous vehicles keep the safety spacing distance. The automobile braking process can be divided into three stages [13]. The ﬁrst one is the reaction and action stage. After the driver faces an emergency scenario until to the vehicle brake, time t1 can be divided into reaction time tr1 and action time ta1 . tr1 is about 0.3–1 s and ta1 is 0.045 s according to the survey. The second one is the deceleration growth stage which spends time t2 . During t2 , the brake force is increased from 0 to the maximum, and the deceleration is gradually increased to maximum. t2 is about 0.025 s. The third stage is the uniform deceleration stage which spends time t3 . The speed of the vehicle keeps uniformly decelerating until vehicle stop. It is assumed the front vehicle A traveling at the constant speed Va and the back one B at Vb . The distance between them is D. Va remains unchanged and Va \Vb .

Research on the Development Status of the Heavy-Duty Truck

373

On this certain situation, the traveling distance of the front vehicle A is Da shown as Eq. (1) and the traveling distance of back vehicle B is Db shown as Eq. (2) Da ¼ Va ðt1 þ t2 þ t3 Þ

ð1Þ

1 1 Db ¼ D1 þ D2 þ D3 ¼ Vb t1 þ Vb t2 ct23 þ Va t3 þ amax t32 6 2

ð2Þ

Deceleration rate is related to amax in time t2 . c¼

da amax ¼ dt t2

ð3Þ

To ensure there is no collision, t3 can be derived. t3 ¼

Vbt2 Va amax

ð4Þ

Combining Eqs. (1)–(4), the traveling distance Da of the front vehicle A can be represented as Eq. (5) and the distance of back vehicle can be represented as Eq. (6). 1 Va Vb Va2 Da ¼ Va ðt1 þ t2 þ t3 Þ ¼ Va ðt1 þ t2 Þ þ 2 amax Db ¼ D1 þ D2 þ D3 ¼ Vb ðt1 þ

1 V 2 Va2 1 t2 Þ þ b amax t22 2 24 2amax

ð5Þ ð6Þ

Without collision, the driving distance can be represented as: D Db Da . Combining Eqs. (5) and (6), the safety spacing distance can be represented as Eq. (7). 1 1 1 D Db Da ðVb Va Þðt1 þ t2 Þ þ ðVb Va Þ2 amax t22 2 2amax 24

ð7Þ

4 Experiment and Simulation To develop the ACC system for the in-wheel motor driving electric truck, the platform of the electric vehicle is established. Based on the designed model, the Omnidirectional steering electric vehicle platform is processed and manufactured. The inwheel motor is replaced by the integrated wheel as shown in Fig. 3(a). The steering wheel is connected with the steering shaft of the steering mechanism, and the four wheels are driven by the steering shaft which is controlled by the steering wheel. The power and rated voltage of each integrated wheel motor are 5.5 kW and 72 V, and each steering motor is 60 W and 15 V. The battery pack is including four batteries of 72 V. Each independent steering mechanism includes a DC motor which uses the gear reducer to reduce speed and increase torque, a driving gear, a group of

374

F. Wei et al.

gear and gear rack which change the rotary motion into linear motion, and a steering rod which connects drive rack and vehicle steering knuckle arm. Thus, each independent steering mechanism drives a correlative in-wheel driving motor independent steering. The ﬁnalized platform is shown in Fig. 3(b) and (c).

Fig. 3. (a) the independent steering mechanism; (b) the ﬁnalized platform; (c) the prototype.

As per experience, the time of the driver’s reaction time and the starting acceleration time are generally set for 1 s, and the unit of velocity is generally for 1 km/h, so the acceleration of the back vehicle can be expressed as Eq. (8). Afv ¼

Vb2 Ve2

S0 þ

Va2 Ve2 2amax

1:0125Vb D

ð8Þ

Ve is the end speed of the two vehicle. S0 is the minimum spacing between the two vehicles. The ACC system model is shown as Fig. 4. And the simulation is set for three travelling condition. The ﬁrst condition is a reduction condition. The speeds of three vehicles are Va ¼ 40 km/h, Vb ¼ 60 km/h and Vc ¼ 80 km/h. The distance of every two contiguous vehicles is 40 m. The purpose of this simulation is mainly to judge if the system can make sure that vehicle C and vehicle B will decelerate accurately and timely. The result is shown in Fig. 5.

Fig. 4. ACC system simulation model.

Research on the Development Status of the Heavy-Duty Truck

375

Fig. 5. (a) is the acceleration change curve of the reduction condition; (b) is the spacing change between the contiguous vehicle of the reduction condition; (c) is speed

The second condition is breaking condition. The speeds of three vehicles are Va ¼ 0 km/h, Vb ¼ 20 km/h and Vc ¼ 40 km/h. The distance of every two contiguous vehicles is 60 m. The purpose of this simulation is mainly to judge if the system can achieve the emergency brake. The result is shown in Fig. 6.

Fig. 6. (a) is the acceleration change when breaking; (b) is the distance between the contiguous vehicle when breaking; (c) is the speed change curve when breaking.

The third condition is the recovery cruise condition. The initial speed of three vehicles is Va ¼ Vb ¼ Vc ¼ 60 km/h. The distance of every two contiguous vehicles is 80 m. The purpose of this simulation is mainly to judge if the system can make the vehicle quickly into the cruise state after vehicle start up. The result is shown in Fig. 7.

Fig. 7. (a) the acceleration change at recovery cruise condition; (b) the distance between the contiguous vehicles at recovery cruise condition; (c) the speed change curve at recovery cruise condition.

376

F. Wei et al.

From the above ﬁgure, it can be concluded that: the designed ACC system can realize to regulate the vehicle speed and adjust the distance under three kinds of travelling conditions, but also from the speed responsibility, it can be seen the fuzzy controller can adapt to different environment after learning and training of neural network fuzzy. The designed system is good to realize the fast response speed and high regulation precision.

5 Conclusion By analyzing the characteristics of vehicle dynamics, this paper designs the electric truck driven by the 4 in-wheel motors, and the Omni-directional steering electric vehicle platform is processed and manufactured. To achieve safe and efﬁcient ACC, safety spacing distance is analyzed. And the ACC system is simulated and veriﬁed on Matlab/Simulink environment. From the simulation results, it can be seen that the good performance of the response proﬁle veriﬁed the designed system over the set travelling condition. The results of this prototype can be presented in a future study. Acknowledgement. This work was supported by the National Key R&D Program of China (2016YFC0802208), the National Natural Science Foundation of China under Grant No.51175443, the Science and Technology Projects of Sichuan under 2014RZ0036, 2015RZ17 and 2016GZ0026, and the Fundamental Research Funds for the Central Universities under Grant No. 2682016ZDPY03.

References 1. Shi, W.M.: World heavy duty vehicle market development forecast. Auto Ind. Res. 5, 14–20 (2001) 2. Gong, Y.N.: Heavy duty vehicle demand analysis and forecast. Automob. Parts 8, 40–44 (2016) 3. Yang, Z.S.: High-end light truck market and products. Automob. Parts 21, 42–45 (2014) 4. Ma, L.: Analysis on the market trend and competition pattern of medium and heavy truck in 2013 and the enterprise plan in 2014. China Auto Sci. Tech. 1, 44–51 (2014) 5. Jing, S.T.: Comparison and analysis of fuel consumption standards for light vehicles. Energy Environ. 2, 40–42 (2008) 6. Wang, W., Li, Y.F.: Fuel consumption limits for freight vehicles. Truck. Logist. 6, 50–52 (2007) 7. Yuan, X.: Latest accident research report: fatigue careless driving induces most truck accidents. Automob. Parts 8, 32–33 (2013) 8. He, Y.T., Wang, Y., Zhu, X.C.: analysis on heavy truck accidents based on in-depth investigation on road trafﬁc. Agric. Equip. Veh. Eng. 8, 1672–3142 (2008) 9. Qin, G., Zou, J., Xu, H., Xin, X., Li, K.: Torque allocation strategy of 4WID in-wheel motor electric vehicle based on objective optimization. In: American Control Conference, pp. 2600–2605. IEEE (2014) 10. Xin, X., Zheng, H., Xu, H., Qin, G.: Control strategies for four in-wheel driven electric vehicles when motor drive systems fail. In: American Control Conference, pp. 885–890. IEEE (2014)

Research on the Development Status of the Heavy-Duty Truck

377

11. Wang, B., Huang, X., Wang, J., Guo, X., Zhu, X.: A robust wheel slip control design for inwheel-motor-driven electric vehicles with hydraulic and regenerative braking systems. In: American Control Conference, pp. 3225–3230. IEEE (2014) 12. Luo, Y.J., Chen, T., Zhang, S., Li, K.: Intelligent hybrid electric vehicle ACC With coordinated control of tracking ability, fuel economy, and ride comfort. IEEE Intell. Transp. Syst. 16, 2303–2308 (2015) 13. Glaser, S., Orﬁla, O., Nouveliere, L., Potarusov, R., Akhegaonkar, S., Holzmann, F., Scheuch, V.: Smart and Green ACC, adaptation of the ACC strategy for electric vehicle with regenerative capacity. In: Intelligent Vehicles Symposium, vol. 36, pp. 970–975. IEEE (2013)

Dynamic Brain Network Evolution in Major Depressive Disorder Liping Yang, Yingjie Liu, Bo Zhang, and Hongbo Liu(B) School of Information Science and Technology, Dalian Maritime University, Dalian 116026, Liaoning, China {liping.yang,lyj,bzhang,lhb}@dlmu.edu.cn

Abstract. Major depressive disorder (MDD) is a common disease of serious mental disorders, and it is characterized by the alterations in brain functional connections. Brain network analysis is contributed to understanding the pathological mechanism of MDD. In this paper, we establish a brain network based on the structural Magnetic Resonance Imaging (sMRI) and present a dynamic evolution model of the brain network. The model uses control factors and evolution strategies to simulate the dynamic process of brain lesions of MDD. The experimental results illustrate that our evolution model captures the dynamics of the MDD, which provides a new way for exploring the pathological mechanism of MDD. Keywords: Major depressive disorder Network evolution

1

· Structural brain network

Introduction

The human brain is known as one of the most complex systems, consisting of billions of neurons that form a hierarchical and highly self-adapting organizational structure [2]. Various bioelectrical signals in the brain are constantly experiencing life processes such as generation, conduction and disappearance. Major depressive disorder (MDD) patients may be abnormal in the physiological structure when bioelectrical signals experience a certain life process, resulting in abnormal function [1,12,13]. However, the dynamic processes and mechanisms of the MDD in the brain remain unclear. Addressing this issue is of great signiﬁcance, not only for a better understanding of normal cognitive depression but also for the early diagnosis and treatment. Therefore, how to establish the structural brain network and explore the pathological process of MDD patients’ structural brain network is important. Graph theoretic analysis provides a new perspective to characterize complex network properties. It has been widely applied to the research of the MDD. The brain is considered as a graph including numerous nodes and edges, which represent brain regions and connectivity between them, respectively [5]. The topological lesions of the structural brain network [7,14] and the functional brain c Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 378–385, 2019. https://doi.org/10.1007/978-3-030-04582-1_44

Dynamic Brain Network Evolution in Major Depressive Disorder

379

network [3,4] have been found. Although researches ﬁnd the changes of static characteristics of MDD patients, a speciﬁc network model can not eﬀectively reﬂect changes of the brain structure, and many dynamic features are ignored. Network evolution has been eﬀectively applied in the research of dynamic characteristics of real system networks [6,8,9]. Some researches use the Euclidean distance as the control factor of network model to simulate the evolution process of brain network [15]. Although network dynamics have been used to explore the brain pathological patterns of autistic patients, the dynamic evolution of MDD has not been well investigated. In this paper, we establish a brain network based on the structural Magnetic Resonance Imaging (sMRI), and compare the structural diﬀerences between the MDD groups and Healthy Controls (HC) groups. Then, the dynamic evolution model is designed to simulate the dynamic process of brain lesions of MDD, which reveals the instantaneous status of the disease trend of the patients with MDD. The model also provides theoretical and practical support for the research on the pathological mechanism of the MDD.

2

Materials and Methods

In this section, we establish the structural brain network based on the gray matter volume to ﬁnd the correlation between the abnormal brain regions. Then, we present a dynamic evolution model using control factors and evolution strategies to simulate the dynamic process of brain lesions of MDD. 2.1

Participants

14 medication-naive participants with MDD aged 19–25 years (6 males, 8 females) and 18 healthy controls aged 19–25 years (9 males, 9 females) were recruited. All participants were recruited from the School of Information at Dalian Maritime University. 2.2

Preprocessing

MRI images were using a 3.0 T Siemens Trio MRI scanner. Acquisition was performed using an 8-channel head coil. Three-dimensional T1-weighted anatomical images were acquired using a 3D SPGR sequence (T R = 1680 ms; T E = 285 ms; F lipAngle = 9; T I = 500 ms; N EX = 1; ASSET = 1.5; Frequency direction: S/I). A total of 160 contiguous 1 mm slices were acquired with a 384×384 matrix, with an in-plane resolution of 1 mm × 1 mm resulting in isotropic voxels. 2.3

Structural Brain Network Establishment

Firstly, we extract the gray matter volume for 90 AAL brain regions, similar to [10], and obtain the gray matter volume of MDD and HC groups, respectively. Secondly, we compute the correlation coeﬃcient of gray matter volume using

380

L. Yang et al.

Pearson correlation coeﬃcient, and obtain two 90 × 90 correlation coeﬃcient matrices. Since the weight of the network is not considered, the obtained relation matrix needs to be converted into a binarization matrix, that is, an adjacency matrix. Finally, we establish the structural brain network, where the brain region is deﬁned as a node and the resulting Pearson correlation coeﬃcient between two brain regions is deﬁned as a link. The whole process of network establishment is shown in Fig. 1.

Fig. 1. General process of establishing the structural brain network

2.4

Evolution Rules

In this section, we take the structural brain network of the HC groups and the MDD groups as the targets, and develop a network evolution model to explore the pathological mechanism of the MDD. Evolution rules are as follows. (1) The starting point for simulating the pathological process is the structural brain network of the HC group. (2) Connections and disconnections between nodes do not occur independently. That is, there is a probability of disconnection of any one of the connected edges in the network and a possibility of connection between any two unrelated nodes. The probability of disconnection and connection is aﬀected by the Euclidean distance and the node degree between two nodes. (3) During the evolution, turbulent factors are required. (4) During the simulation, the generation and disappearance of the edge are alternated.

Dynamic Brain Network Evolution in Major Depressive Disorder

381

Considering the inﬂuence of the Euclidean distance and the node degree on information transmission, we set the following disconnection probability (Eq. 1) and edge probability (Eq. 2) for each edge. p dis(i, j) = D(i, j)x · (di + dj )−y ;

(1)

p dis(i, j) = D(i, j)−x · (di + dj )y .

(2)

where x, y are the evolutionary control factors. D(i, j) represents the Euclidean distance between i and j, di and dj represent the degree of i and j, respectively. Among them, the probability of connection between any two nodes is directly proportional to the degree of node and inversely proportional to the Euclidean distance. The disconnection probability of any edge is proportional to the Euclidean distance, and is inversely proportional to the degree of the node. In the process of evolution, we join a turbulent factor R. R is a random matrix generated randomly from elements 0 to 1. The speciﬁc operation is listed as follows: (1) If there is no edge between nodes and pcon (i, j) > 0.5, R < 0.03, a new connection is established between nodes. (2) If there is a edge between nodes and pdis (i, j) > 0.5, R < 0.03, the nodes are disconnected. (3) In other cases, the status between nodes remains unchanged. In order to simulate the pathological process of the brain, it is necessary to ﬁnd the optimized x and y. We use the diﬀerential evolution algorithm to ﬁnd x and y, proposed by Storn and Price [11]. The cost function is deﬁned in Eq. 3. 90 90 |C(i)w − C(i)dep | Ew − Edep n=1 |D(i)w − D(i)dep | · · n=190 . (3) Error = 90 Edep n=1 D(i)dep n=1 C(i)dep where Ew represents the global eﬃciency of the evolution network W , Edep represents the global eﬃciency of the MDD network, D(i)w and D(i)dep represent the degree distribution of the i node in the simulation network W and the MDD network, respectively, C(i)w and C(i)dep represent the clustering coeﬃcient of the i node in the simulation network W and the MDD network, respectively. The ultimate goal of simulating the pathological process is to minimize the error. The framework of the brain network evolution is shown in Fig. 2.

3

Experiments and Results

In this section, we analyze topology properties and structural diﬀerences structural brain networks. Then, we compute and compare various topological properties of HC group network, MDD network and evolution network W , accordingly. 3.1

Structural Brain Network

We set the threshold value by checking whether the network has small world characteristics. The threshold is set to 0.7, and the resulting brain network is shown in Fig. 3, where the left represents the HC group and the right represents the MDD group.

382

L. Yang et al.

Fig. 2. Framework of the dynamic brain network evolution

Fig. 3. Structural brain network of two groups

The density of the edge of the HC group is bigger than that of the MDD group. The HC group was closely associated with other brain regions in the ACG.L, while the MDD group was not connected to other brain regions. Secondly, the density of connections between the brain regions of the HC group is relatively larger in the ORB, while the MDD group is sparse. The lower part of the ORBinf.L and the middle of the MTG.L is an isolated node in the network of the MDD group. These results indicate that the internal structure of the brain network of the MDD group is signiﬁcantly diﬀerent from the HC group.

Dynamic Brain Network Evolution in Major Depressive Disorder

3.2

383

Evolution Results

In order to determine the evolution steps of the network, we select the ﬁxed x and y. It is found that the average clustering coeﬃcient and the global eﬃciency of the network tend to be stable when the evolution is around 200 steps, that is, it converges to the ﬁxed value. Evolution process is shown in Fig. 4. Under the control of DE, the evolution error is less than 0.5. We set 6 diﬀerent thresholds to obtain 6 sets of binarized networks. The values of 6 groups of x and y are shown in Table 1. Each group includes the structural brain network of MDD and HC groups, and the HC group network is used as the starting point to simulate the pathological process. The error computed by the DE algorithm is shown in Fig. 5.

Fig. 4. Brain network evolution curve

Fig. 5. Objective function curve of DE

Table 1. Evolutionary control factor under diﬀerent threshold Threshold 0.60

0.62

0.64

0.66

0.68

0.70

x

2.7084 2.4686 2.4571 2.5861 2.1394 2.2043

y

2.8851 2.7170 2.7829 2.9908 2.5571 2.7318

In order to measure the eﬀectiveness of evolution, we select two groups of structural brain networks under diﬀerent thresholds and simulate the pathological process of MDD, and compare the edge number, global eﬃciency and other topological attributes of the three networks (HC group network, MDD network, evolution network W ). As shown in Fig. 6, the number of edges of the evolution network W under multiple threshold conditions has converged to MDD group network. The global eﬃciency and the transitivity of the evolution network W are also close to the MDD network. The value of feature path length of the evolution network W is always higher than that of the HC group network. It is veriﬁed that the feature path length is inversely proportional to the information transmission eﬃciency, and the tendency of the curve increases with the increase

384

L. Yang et al.

(a) Edges

(b) Global-Efficiency

(c) Transitivity

(d) Characteristic-Path-Length

Fig. 6. Topological property in the three kinds networks under multiple threshold

of the threshold. It is found that the number of edges of the simulation network W is almost the same as that of the MDD network. Therefore, the internal structure of the simulation network generated by the evolution model is similar to the structural brain network of the MDD.

4

Conclusion

In this paper, we establish a structural brain network using MRI and compare the structural diﬀerences between the MDD and the HC groups. Then, the dynamic evolution model is designed to simulate the dynamic process of brain lesions of MDD. As a result, the internal structure of the simulation network generated by the evolution model is similar to the structural brain network of the MDD. The experimental results verify the reliability and eﬀectiveness of the simulated MDD pathology model. Our model is of great signiﬁcance for future explorations of the mechanisms of the MDD. Acknowledgment. The authors would like to thank Qianrui Qi for his scientiﬁc collaboration in this research work. This work is partly supported by the National Natural Science Foundation of China (Grant No. 61472058, 61702073, 61751205, 61602086 and 61772102).

Dynamic Brain Network Evolution in Major Depressive Disorder

385

References 1. Bora, E., Fornito, A., Pantelis, C., Y¨ ucel, M.: Gray matter abnormalities in major depressive disorder: a meta-analysis of voxel based morphometry studies. J. Aﬀect. Disord. 138(1–2), 9–18 (2012) 2. Carhart-Harris, R.L., Friston, K.J.: The default-mode, ego-functions and freeenergy: a neurobiological account of freudian ideas. Brain 133(4), 1265–1283 (2010) 3. Chen, J.E., Glover, G.H.: Functional magnetic resonance imaging methods. Neuropsychol. Rev. 25(3), 289–313 (2015) 4. Chen, L., Zhang, W., Liu, H., Feng, S., Chen, C.P., Wang, H.: A space aﬃne matching approach to fmri time series analysis. IEEE Trans. Nanobiosci. 15(5), 468–480 (2016) 5. Fallani, F.D.V., Astolﬁ, L., Cincotti, F., Mattia, D., Marciani, M.G., Salinari, S., Kurths, J., Gao, S., Cichocki, A., Colosimo, A., et al.: Cortical functional connectivity networks in normal and spinal cord injured patients: evaluation by graph analysis. Hum. Brain Mapp. 28(12), 1334–1346 (2007) 6. Han, Q., Yu, L., Zheng, W., Cheng, N., Niu, X.: A novel QKD network routing algorithm based on optical-path-switching. J. Inf. Hiding Multimed. Signal Process. 5(1), 13–19 (2014) 7. Le Bihan, D.: Looking into the functional architecture of the brain with diﬀusion MRI. Nat. Rev. Neurosci. 4(6), 469 (2003) 8. Li, X., Jin, Y.Y., Chen, G.: Complexity and synchronization of the world trade web. Phys. A Stat. Mech. Appl. 328(1–2), 287–296 (2003) 9. Pagani, G.A., Aiello, M.: Power grid complex network evolutions for the smart grid. Phys. A Stat. Mech. Appl. 396, 248–266 (2014) 10. Qi, Q., Wang, W., Deng, Z., Weng, W., Feng, S., Li, D., Wu, Z., Liu, H.: Gray matter volume abnormalities in the reward system in ﬁrst-episode patients with major depressive disorder, pp. 704–714. Springer (2018) 11. Qiu, X., Xu, J.X., Xu, Y., Tan, K.C.: A new diﬀerential evolution algorithm for minimax optimization in robust design. IEEE Trans. Cybern. PP(99), 1–14 (2018) 12. Sara, P., Veronica, A., Silvia, B., Irene, B., Andrea, F., Cristina, C., Francesco, B.: Impact of early and recent stress on white matter microstructure in major depressive disorder. J. Aﬀect. Disord. 225, 289–297 (2018) 13. Sawyer, K., Corsentino, E., Sachs-Ericsson, N., Steﬀens, D.C.: Depression, hippocampal volume changes, and cognitive decline in a clinical sample of older depressed outpatients and non-depressed controls. Aging Ment. Health 16(6), 753– 762 (2012) 14. Tang, E., Giusti, C., Baum, G.L., Gu, S., Pollock, E., Kahn, A.E., Roalf, D.R., Moore, T.M., Ruparel, K., Gur, R.C., et al.: Developmental increases in white matter network controllability support a growing diversity of brain dynamics. Nat. Commun. 8(1), 1252 (2017) 15. V´ertes, P.E., Alexander-Bloch, A.F., Gogtay, N., Giedd, J.N., Rapoport, J.L., Bullmore, E.T.: Simple models of human brain functional networks. Proc, Natl. Acad. Sci. (2012). https://doi.org/10.1073/pnas.1111738109

Passenger Centric Timetable Synchronization in Metro Network Yuanyuan Wang(&) School of Business Administration, Zhejiang University of Finance and Economics, Hangzhou 310018, China [email protected]

Abstract. Timetable synchronization is an effective measure to improve the service level of metro network. This paper proposes a timetable synchronization optimization model from the perspective of passengers, what we called Passenger Centric Timetable Synchronization (PCTS) model. The model aims to maximize the satisfaction of passengers by adjusting the departure time, running time, dwelling time and headways for all transfer pairs in the metro network. Genetic algorithm is applied to solve the model. Finally, the timetable of Hangzhou metro network is optimized to verify the effectiveness of the above model. The results show that the PCTS model can be used to improve the satisfaction of passengers in metro network signiﬁcantly. Keywords: Passenger centric

Timetable synchronization Metro network

1 Introduction On the background of trafﬁc congestion, air pollution and energy crisis, metro system plays a vital role in the sustainable development of cities. Passengers usually make several transfers during their travel. Mohring et al. (1987) found that passengers often view their waiting time to be twice of what it actually is. Timetable synchronization in the network is an effective measure to reduce transfer waiting time. Timetable is the main product of metro operation company. Decades of efforts have been invested by the researchers to generate the coordinated timetable with the objective of minimizing the total transfer waiting time (Domschke 1989; Wong et al. 2008). Although some Metro Operation Companies(MOC) have designed the coordinated timetable in practice aiming to minimize the total waiting time in the network, they still received a lot of complaints from passengers about the long transfer waiting time. The reason is that minimizing the overall waiting time of the network is an aggregate result, and it doesn’t mean the satisfaction of individual passenger. Since the consumers of MOC are passengers, it is important to account for the passengers’ preferences during timetable synchronization. There is a wealth of literatures on timetable synchronization. Most models were to minimize the total transfer waiting time or waiting cost. Wong et al. (2008) propose a mixed integer programming optimization model to minimize interchange waiting time of all passengers in a railway system. A novelty in their formulation is the use of binary variables which enable the correct representation of the waiting-times for transfer to the © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 386–393, 2019. https://doi.org/10.1007/978-3-030-04582-1_45

Passenger Centric Timetable Synchronization in Metro Network

387

“next available” train at interchange stations. Wu et al. (2015) proposed a timetable synchronization optimization model to optimize passengers’ waiting time while limiting the waiting time equitably over all transfer station in an urban subway network. Another objective in the previous modes is to maximize the synchronization. Ceder et al. (2001) propose a mixed integer linear model to maximize the synchronization in the bus network. Eranki (2004) extended the model in Ceder et al. (2001) by deﬁning the ‘simultaneous arrival’ as two arrivals of buses of different routes do not exceed the passenger waiting time range associated with the transfer stop. We propose a passenger satisfaction function, and construct a passenger’s satisfaction-based timetable synchronization model with the departure time from the starting station, dwelling time and running time as decision variables. The structure of this paper is organized as follows: Sect. 2 explains the details of the objective function and mathematical model. Section 3 presents a heuristic algorithm to solve the model. In Sect. 4, a numerical example is given to verify the effectiveness of the suggested model.

2 Model Formulation 2.1

Assumption

The following assumptions are made in the paper: The number of transfer passengers in each transfer direction is known and ﬁxed; passengers have an average walk speed, therefore, the transfer time for each transfer pair is assumed to be a constant value; the capacity of the trains is enough at any time, and all the transfer passengers can enter into the next possible connecting train successfully; since there is no transfer activity in the intermediate station, we only consider the transfer and terminal stations in our study; the dwelling time of trains on the intermediate stations are included in the running time between two adjacent transfer stations; Since we study only one period, such as morning rush hour or evening rush hour, we assume that the passengers are homogeneous, or we assume the passengers studied here are all commuters. 2.2

Notation

L: Set of lines in the network, L ¼ fl=l ¼ 1; 2; . . .; ng, where n is the total number of lines in metro network. We view the line with the up and down directions as two l individual lines; Sl ¼ fs1l ; s2l ; . . .sm l g: Set of terminal and transfer stations of line l.ml is the total number of transfer stations of line; Tl ¼ ftl1 ; tl2 ; . . .tlrl g: Set of trains scheduled 0 on line l. rl is the total number of trains on line l; U: Set of transfer pairs, ðl; l ; jÞ 2 U, 0

0

which means passengers transfer from line l to line l at transfer station j; ell;j : Transfer 0

0 0

time for passengers transfering from line l to line l at station j; cljltt : Number of 0 passengers who transfer from the tth train on line l to the tth train on line l0 at transfer 0 0 station j; hl : Headway of line l; wljltt : Waiting time of passengers who transfer from the 0 0

0 train on line l0 at transfer station j. aljltt ¼1: If the tth train on tth train on line l to the tth 0 line l arrives early enough that passengers can transfer to the tth train on line l0 at

388

Y. Wang 0 0

transfer station j successfully; otherwise, aljltt ¼0; Atlj : Arrival time of the tth train at station j on line l; Ltlj : Departure time of the tth train from station j on line l, j 6¼ 1. Decision Variables Ltl1 : Departure time of the tth train from the starting station of line l; Rtlj : Running time of the tth train from station j1 to station j on line l; Dtlj : Dwell time of the tth train at station j on line l. 2.3

Model Formulation

2.3.1 Objective Function We set four satisfaction levels: those are Satisﬁed, General, Dissatisﬁed and Dissatisﬁed extremely. We conducted a SP questionnaire survey on Internet, the key content of which is that “how long is the waiting time which you’re satisﬁed/dissatisﬁed/ dissatisﬁed extremely when the headway of the connecting line is hl minutes?” We 0 0 deﬁne passenger satisfaction degree Wðwljltt Þ, which is the function of waiting time, satisfaction threshold and headway we found in a survey. We deﬁned satisfaction index ui be the ratio of satisfaction threshold to headway for each kind of satisfaction level. We assign the numerical values l1 , l2 , l3 and l4 to each of the satisfaction level respectively. The satisfaction level function can be formulated as Eq. (1). 8 l1 > > > l3 > > : l4

0 0

0 wljltt u1 hl0 0 0 u1 hl0 \wljltt u2 hl0 0 0 u2 hl0 \wljltt u3 hl0 0 0 u3 \wljltt hl0

ð1Þ

The objective function is to maximize passenger satisfaction degree in the metro network, as shown in Eq. (2). max F ¼

rl0 rl X X X 0

ðl;l ;jÞ2U t¼1

t0 ¼1

0 0

0 0

Wðwljltt Þ cljltt

ð2Þ

2.3.2 Constraints For the train t on the line l 2 L, the arrival time Atlj and the departure time Ltlj for train t at/from the transfer station j can be deduced by Eqs. (3) and (4). Atlj ¼ Ltl1 þ

j X i¼1

Rtli þ

Ltlj ¼ Atlj þ Dtlj

j1 X i¼1

Dtli

ð3Þ ð4Þ

Passenger Centric Timetable Synchronization in Metro Network

389

l

In Eqs. (5) and (6), d lj and d j are the minimum and the maximum dwelling time at l

station j on line l. Rlj and Rj are the minimum and the maximum running time between the transfer station j − 1 and j on line l. d lj Dtlj d j

l

ð5Þ

l

ð6Þ

Rlj Rtlj Rj

For the two adjacent trains t–1 and t on the line l, the time interval between their arrival time and departure time at/from the station j should satisfy the following constraint, as shown in Eq. (7) and (8). hj Atlj At1;l hj j

ð7Þ

hj Ltlj Lt1;l hj j

ð8Þ

The calculation of waiting time is as proposed in Wong et al. (2008). The deﬁnition of transfer waiting time is shown in Fig. 1. If passengers from the tth feeder train can’t 0 0 get on the (t0 1)th connecting train, the transfer waiting time wljltt 1 is enforced to be 0 0

0 0

zero, as well as the waiting time wljltt 2 and wljltt 3 . The model considers only the transfer waiting time for passengers getting on the next possible connecting train. Feeder train t wljlt't '-1 =0 l ' t '-2 jlt

w

wljlt't '+1 =0

=0

wljlt't '-3 =0

Arrival time

e

l' j ,l

wljlt't ' > 0

time

Departure time Connecting train t '− 3

t '− 2

t '− 1

t'

t '+ 1

Fig. 1. Deﬁnition of the transfer waiting time (Wong et al. (2008)) 0 0

The binary variable aljltt ¼1 represents the tth train on line l arrives early enough that passengers can transfer to the t0 th train on line l0 at transfer station j successfully. 0 0 Otherwise, the transfer is unsuccessfully. aljltt can be obtained by the Eq. (9). 0 0

0

0 0

0 0

l t 1 wljltt Ltj l ðAtlj þ elj;l Þ Majlt

ð9Þ

390

Y. Wang 0 0

The waiting time wljltt can be obtained by Eq. (11). M is an large positive number. 0 0

Note that the waiting time wljltt is always smaller than the headway hl0 and larger than 0, as shown in Eq. (11). 0 0

0 0

0 0

Mðaljltt 1Þ Ltj l ðAtlj þ elj;l0 Þ Maljltt 0 0

0 wljltt hl0

ð10Þ ð11Þ

3 Solution Approach Genetic algorithm is a heuristic algorithm inspired by the process of natural selection, and can solve large scale optimization problem efﬁciently. We use genetic algorithm to solve the above model. A chromosome is composed of the decision variables, including departure time from the starting station Ltl1 , running time on each section Rtlj and rn n dwelling time Dtlj . Therefore, the chromosome can be expressed as ðL11 1 ; . . .; L1 ; 11 rn n 11 rn n R1 ; . . .; Rml ; D1 ; . . .; Dml Þ. The ﬁrst generation is created randomly. The objective function F is chosen to be the ﬁtness function, that is Eq. (2). The genetic operators we adopt here are reproduce, crossover and mutation. First, the individuals are selected to reproduce themselves for each generation. And then, individuals are paired up to execute the crossover operator with a speciﬁc rule. Thirdly, the mutation operator is executed when the mutation points are selected according to the mutation fraction we set. At last, the ﬁtness of each individual is calculated, and we choose the best individual in each generation. For a faster convergence in the algorithm process, we adopt the elitism strategy, which is achieved by substituting the worst 4% chromosomes in the offspring with the best 4% in the parents. During the algorithm, we verify the feasibility of each individual after each operator, if not, we will repeat the operator until the individual is feasible.

4 Case Study 4.1

Parameters

There are three two-way lines and 84 stations in Hangzhou Metro network. We view the two-way line as two individual lines. Therefore, we focus on six lines, ﬁve transfer stations and four terminal stations. For description conveniently, we assign a serial number to each station, as shown in Fig. 2. The running time between two adjacent stations are shown in Table 1. The dwelling time at Station 3 and 7 is set to be 30 s, and the others are 25 s. The numbers of trains scheduled on the Line 1, 2 … and 6 are 15, 15, 10, 10, 12 and 12 respectively. The headway of each line from line 1 to line 6 are 240, 240, 360, 360, 300 and 300 respectively. The allowable adjustments of running time, headway and dwell time are 10%, 30 s and 10% respectively. The transfer pair, the transfer time and the transfer passengers demand in each transfer pair are shown in Table 2.

Passenger Centric Timetable Synchronization in Metro Network

Fig. 2. Hangzhou metro network

Table 1. Running time in each section (Unit: sec) Section Station 12 Station 23 Station 27 Station 34 Station 37

Running time Section Running time 840 Station 38 840 780 Station 45 180 540 Station 47 1080 840 Station 56 1800 720 Station 79 1800

Table 2. Transfer time and transfer passengers in each transfer pair Station Transfer pair 2 Line 1->3 Line 2->3 3 Line 1->5 Line 1->6 Line 2->5 Line 2->6 4 Line 1->6 Line 2->6 5 Line 3->1 7 Line 6->4 Line 6->3

Transfer time Transfer (sec) passengers 30 90 10 65 25 86 15 42 25 85 15 30 20 99 10 10 25 108 10 68 20 75

Transfer pair Line4->1 Line4->2 Line5->1 Line5->2 Line6->1 Line6->2 Line5->1 Line5->2 Line2->4 Line5->3 Line5->4

Transfer time (sec) 30 10 25 15 25 15 20 10 15 10 20

Transfer passengers 62 22 90 30 20 15 10 89 98 58 10

391

392

Y. Wang

By a SP questionnaire survey, the result of which is shown in Table 3, we got the parameters u1 ,u2 and u3 are 0.31, 0.49 and 0.85 respectively. For the numerical values of each satisfaction level, we set l1 ¼2, l2 ¼1, l3 ¼0 and l4 ¼ 1. We run the genetic algorithm program in MATLAB (7.0). We set the population size to be 300. The length of chromosome in this case study is 962. The crossover fraction and mutation fraction are 0.8 and 0.1 respectively. We set the max generation to be 200.

Table 3. Data of the questionnaire survey Satisfaction level

Dissatisfaction extremely Headway (min) 6 10 15 6 10 15 6 10 15 Average threshold (min) 1.93 3.02 4.55 3.00 4.89 7.48 4.90 8.53 13.18 Satisfaction index 0.31 0.49 0.85

4.2

Satisfaction

Dissatisfaction

Result Analysis

The timetable synchronization result of Hangzhou Metro network is shown in Table 4. We can see that, although the total waiting time increased by 24.2% in our model, the passenger satisfaction increased by 19.9%. And, the number of passengers who are satisﬁed increased by 25.1%. It proved that the model we propose can improve the passenger satisfaction effectively.

Table 4. Result of the case study Timetable Total waiting time

Passenger satisfaction

Previous 785 Optimized 975

1609 1930

Improvement Number of passengers Satisﬁed Dissatisﬁed Dissatisﬁed Extremely – 689 287 230 19.9% 862 174 10

5 Conclusion This paper presents a Passenger Centric Timetable Synchronization (PCTS) model to optimize the timetable of metro network with the objective of maximizing passenger satisfation. We deﬁne passenger satisfaction degree parameters to express different service level, those are Satisﬁed, General, Dissatisﬁed and Dissatisﬁed extremely. Then, we propose a passenger satisfaction-based model for timetable synchronization with the departure time, running time and dwelling time as decision variables. The optimal result is obtained by genetic algorithm. Our case study of Hangzhou metro network shows that the model we proposed is easy to implement and able to achieve the aim of maximizing the passengers satisfaction compared with the previous waiting time-based model in the literature. The passengers are assumed to be homogeneous in

Passenger Centric Timetable Synchronization in Metro Network

393

our paper, however, the passengers are heterogeneous usually. Therefore, the expression of passenger satisfaction function when the passengers are heterogeneous can be studied in the future research. Funding Statement. This paper is supported by Natural Science Foundation of Zhejiang Province, China (LQ18G030012) and Humanities and Social Science Foundation of Ministry of Education of China (18YJC630190).

References Ceder, A., Golany, B., Tal, O.: Creating bus timetables with maximal synchronization. Transp. Res. Part A Policy Pract. 35, 913–928 (2001) Anitha, E.: A model to create bus timetables to attain maximum synchronization considering waiting times at transfer stops. University of South Florida (2004) Wong, R.C.W., Yuen, T.W.Y., Fung, K.W., et al.: Optimizing timetable synchronization for rail mass transit. Transp. Sci. 42(1), 57–69 (2008) Wu, J., Liu, M., Sun, H., Li, T., Gao, Z., Wang, D.Z.W.: Equity-based timetable synchronization optimization in urban subway network. Transp. Res. Part C: Emerg. Technol. 51, 1–18 (2015) Mohring, H., Schroeter, J., Wiboonchutikula, P.: The values of waiting time, traved time, and a seat on the bus. Staff Gen. Res. Pap. Arch. 18, 40–56 (1987) Domschke, W.: Schedule synchronization for public transit networks. OR Spektrum 11, 17–24 (1989)

Brain Structural Abnormalities in Reward and Emotion System in Internet Addiction Disorder Jinqing Yang1 , Wei Wang2 , Zhongyuan Cao1 , Zhaobin Deng3 , Wencai Weng3 , Shigang Feng1(B) , Hongbo Liu1 , and Mingyu Lu1 1

2

School of Information, Dalian Maritime University, Dalian, China [email protected],{czy,shgfeng,lhb,lumingyu}@dlmu.edu.cn Physical Science and Technical College, Dalian University, Dalian, China [email protected] 3 Aﬃliated Xinhua Hospital, Dalian University, Dalian, China [email protected],[email protected]

Abstract. Internet addiction disorder (IAD) is increasingly recognized as a mental health disorder, particularly among adolescents. The pathogenesis associated with IAD, however, remains unclear. This study aims to explore the changed structural characteristics of IAD adolescents at rest. We recruited 11 adolescents who meet the criteria as IAD group and 11 age and gender matched university students seldom use Internet as Healthy control (HC) group. High-resolution T1- weighted magnetic resonance imaging scans were performed on the two groups. Voxel-based morphometry (VBM) analysis was used to compare the gray matter density (GMD) between the two groups. In addition, a correlation analysis was performed to assess the relationships between GMD of abnormal brain regions and clinical internet addiction measure. Compared with healthy controls, IAD had lower GMD in the amygdala, middle temporal gyrus (MTG), middle occipital gyrus (MOG), orbital frontal cortex (OFC), middle frontal cortex (MFC), anterior cingulate cortex (ACC), thalamus, olfactory, putamen and caudate and the GMD of thalamus has a negative correlation with the Internet Addiction Test (IAT) scores. Our ﬁndings suggested that brain structural changes of IAD present weakened characteristics of IAD in two systems: reward system and emotion regulation system, and this ﬁnding may provide a new insight into the pathogenesis of IAD. Keywords: Internet addiction disorder Gray matter density

1

· Voxel-based morphometry

Introduction

Internet addiction disorder (IAD) is a rapidly growing concern worldwide and it is increasingly recognized as a highly prevalent morbid condition. Psychiatric comorbidities including depression, attention-deﬁcit/hyperactivity disorder, and c Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 394–401, 2019. https://doi.org/10.1007/978-3-030-04582-1_46

Brain Structural Abnormalities in Reward and Emotion System in IAD

395

substance use disorders have been reported in relation to internet addiction, as well as serious health consequences such as cardiopulmonary death [4]. Unlike substance use disorders, no chemical or substance intake is involved in IAD, although excessive Internet use may lead to physical dependence, similar to other addictions [8]. This kind of behavior addiction is deﬁned as an uncontrollable desire for going online, accompanied by devaluation of time spent without the net, nervousness and aggression in situations where Internet is not accessible, and progressive disruption of family and social life [2]. For these reasons, the problem of IAD has attracted much attention from psychiatrists, educators, and the public alike. Especially internet addiction among adolescents, has become an important public concern and gained more and more attention internationally. However, due to the lack of enough comprehensive ﬁndings, the neural mechanism underlying IAD is still not clear and little is known about the treatment of IAD [5]. With the current lack of evidence-based treatments for internet addiction, neurobiological studies may be able to inform their novel development. Resting state is a method of functional brain imaging that evaluates regional interactions that occur when a subject is not performing an explicit task. The resting state approach is useful to explore the brain’s functional organization and to examine if it alters in neurological or psychiatric disease [14]. There are many neuroimaging studies on the changes of brain structure and function in resting state of IAD, and the study of resting-state gray matter alteration could reveal the underlying pathological mechanism of Internet addiction. Structural neuroimaging of brain could be used to investigate brain mechanisms about individual personality traits [9]. The brain’s gray matter is a major component of the central nervous system made up of neuronal cell bodies and it is involved in motor control, perception, memory, emotions, and speech [14]. Voxelbased morphometry (VBM) is a fully automated alternative to the techniques that require volumetric samples to detect the gray matter diﬀerences between groups, and is widely used as an imaging tool to evaluate patterns of brain anatomy change. It is the most commonly applied technique in detecting the morphological substrates. Compared with traditional morphometric approaches which rely on measuring brain volumes manually, VBM is a time-saving technique. In addition, it is not speciﬁc to particular brain regions. This measure was developed to detect group diﬀerences in the relative concentration of gray matter tissues across the whole brain in a voxel manner [17]. The aim of this study is to detect the possible brain morphology changes in IAD adolescents and its mechanism use VBM.

2 2.1

Materials and Methods Subjects

Participants were university students and were recruited through advertisements. Participants were right-handed and were divided into internet addiction disorder group (IAD, 9 males, 9 females) and healthy control group (HC, 9 males, 9 females). IAD and HC groups did not signiﬁcantly diﬀer in age (IAD: mean±SD,

396

J. Yang et al.

19.45 ± 1.37 years; HC: mean ± SD, 20.45 ± 1.69 years; t = 1.52, p = 0.14). All subjects participated in a resting-state magnetic resonance scan. All participants provided written informed consent and all participants were medication free and were instructed not to use any substances, including coﬀee, on the day of scanning. No participants reported previous experience with illicit drugs. IAD was determined based on Young’s online Internet Addiction Test (IAT) scores of 50 or higher (In this study, IAT score in IAD group: mean ± SD, 66.33±8.39; IAT score in HC: mean±SD, 31.06±4.81). Young’s IAT consists of 20 items associated with online Internet use including psychological dependence, compulsive use, withdrawal, related problems in school or work, sleep, family or time management. The IAT has demonstrated validity and reliability and may be used in classifying IAD. For each item, a graded response is selected from 1 = “Rarely” to 5 =“Always”, or “Does not Apply”. Scores over 50 indicate occasional or frequent Internet-related problems. 2.2

Functional Imaging Data Acquisition

Magnetic resonance imaging data were acquired using a Siemens Tro 3.0 T scanner (Siemens, Erlangen, Germany) in Dalian Xin Hua Hospital. Participants lay supinely with the head ﬁxed snugly by a belt and foam pads to minimize head movement. Structural images covering the whole brain were collected by using a T1-weighted three-dimensional spoiled gradient-recalled sequence (160 slices, TR = 1680 ms, slice thickness = 1.0 mm, skip = 0.5 mm, ﬂip angle = 9◦ , ﬁeld of view = 240 × 240, in-plane resolution = 384 × 384). The whole scanning process was a resting state scan lasts for 6 min. 2.3

Data Analysis

Imaging analysis was preprocessed and analyzed using Statistical Parametric Mapping SPM8 (http://www.ﬁl.ion.ucl.ac.uk/spm/software/spm8/). Images were slice-timed, reoriented, and realigned to the ﬁrst volume. The images from all participants were normalized to the Montreal Neurological Institute (MNI) template resulting in an isometric voxel size of 3 × 3 × 3 mm and were segmented into gray matter, white matter and cerebral spinal ﬂuid (CSF). As well as spatial smoothing with a 8 mm full width at half maximum (FWHM) Gaussian kernel. In this study, we only focused on the gray matter destiny (GMD). The analysis of VBM data was focused on the eﬀects of IAD by contrasting the GMD diﬀerences between participants with and without IAD. Voxel-wise comparisons of GMD were performed between two groups using the two-sample t-test with SPM8. The signiﬁcance of group diﬀerences was estimated by the theory of random Gaussian ﬁelds, and signiﬁcance levels were set at *p < 0.01, while the cluster size was set at > 60 voxels.

3

Result

In the present study, VBM analysis indicated decreased gray matter density in several clusters of the IAD compared with the HC, i.e. amygdala, middle

Brain Structural Abnormalities in Reward and Emotion System in IAD

397

Table 1. Regions that show decreased GMD in IAD compared with HC Hemisphere T -value Voxels Talairach coordinates x y z

Regions Amygdala

L

2.57

136

–22.5 2

–19

Middle frontal cortex

L

3.43

107

–43.5 49.5

12

Middle temporal gyrus

L

4.24

117

–45

–6

–16.5

Orbital frontal cortex

R

3.64

101

27

31.5

–9

Putamen

R

2.99

216

24

9

–4.5

Caudate

L

3.26

110

–4.5

9

–1.5

Thalamus

R

4.14

291

19.5

–19.5 13.5

Anterior cingulate cortex R

2.91

60

13.5

45

12 –18

Olfactory

R

4.89

173

16.5

12

Middle occipital gyrus

R

4.15

307

31.5

–79.5 18

temporal gyrus (MTG), middle occipital gyrus (MOG), orbital frontal cortex (OFC), middle frontal cortex (MFC), anterior cingulate cortex (ACC), thalamus, olfactory, putamen and caudate (Table 1 and Fig. 1). We analyzed the correlation between GMD and the IAT scores of all participants, and found negative correlation between GMD of thalamus and the IAT scores (Fig. 2).

Fig. 1. Regions of decreased GMD in IAD compared with HC

4

Discussion

The reward system, also known as the limbic dopamine reward system, is a neural network consisting of the nucleus accumbens, caudate nucleus, putamen,

398

J. Yang et al.

thalamus, hypothalamus, amygdala and other deep brain nuclei and medial prefrontal lobe. Its function is processing stimulus related to reward or expectations of reward. The brain structures that compose the reward system are located primarily within the cortico-basal ganglia-thalamo-cortical loop. In this study, we found decreased GMD in several brain regions of the IAD group in reward system. The OFC has extensive connections with the striatum and limbic regions, and is well situated to integrate the activity of several limbic and subcortical areas associated with motivational behavior and reward processing [16]. It has been reported that IAD has decreased OFC gray matter density and disrupted functional connectivity of OFC-striatum circuit [5]. OFC plays a key role in the reward anticipation, such as sensory rewards and abstract rewards. And it forms a system that controls and manages the cognitive processes in people’s daily lives, with the dorsolateral prefrontal cortex (DLPFC), and the anterior cingulate cortex (ACC) of the prefrontal regions. This system has a major impact on the ability to perform such tasks as planning, prioritizing, organizing, paying attention to and remembering details, and controlling emotional reactions [15]. The impairment of these regions can lead to disinhibition, impulsivity, and has negative eﬀects on individual’s cognitive control of goal-directed behavior so that IAD could not take the correct action to achieve a certain goal based on the information given by the surrounding environment [13].

Fig. 2. Anatomical association with IAT. (A) GMD was negatively correlated with individual IAT scores in the left Caudate (r = 0.37, p < 0.05). (B) GMD was negatively correlated with individual IAT scores in the right OFC (r = 0.35, p < 0.05). (C) GMD was negatively correlated with individual IAT scores in the right Thalamus (r = 0.35, p < 0.05).

It is well known that the forehead area is important in inhibition control, but in fact, these forehead areas do not perform their functions independently, but

Brain Structural Abnormalities in Reward and Emotion System in IAD

399

are often associated with cortical motor areas and basal structures, especially with the dorsal striatum consists of putamen and caudate nucleus. The dorsal striatum is the input nucleus of the basal ganglia and mediates action learning and performance and aspects of cognition. The caudate receives inputs from the associative cortices, and the putamen, interconnecting the cerebral cortex and subcortical areas, eﬃciently transmitting and processing emotional and rewarding information, is the core of the reward network. The anterior putamen is thought to be specialized in goal-directed control or response-monitoring in connection with frontal regions whereas the posterior part is specialized in habitual or automatic responding in connection with sensorimotor regions [1]. Structural changes of this area in IAD we found is consistent with previous studies [12], and may be responsible for patients’ addictive behavior. The thalamus, a subcortical region involved in the brain reward system, is a relay station for the input information in the cerebral cortex, actively sending message to the cortex in various mechanisms [10]. It plays an important role in Cortico-striato-thalamo-cortical circuit (CSTC) [1]. The alteration of thalamus can lead to disorders in the conduction and integration of subcortical emotion and cognitive information. Functionally, repeated exposure to salient stimuli like video games aﬀects dopaminergic pathways and decreases sensitivity to the natural stimuli, resulting in reward processing dysfunction. All those alterations in reward system may be due to a long-term contact to the network. Emotion regulation is another part that we found may be impaired in IAD. The amygdala is an important structure of the limbic system and is the core substrate of emotional processing. It is mainly responsible for the management and control of emotions. It plays a very important role in the withdrawal/negative emotion stage of addiction behavior. Ko and colleagues [7] ﬁrst found that adolescents with Internet addiction had a lower GMD over the bilateral amygdala, their study suggested that the eﬀects of the addiction process, or the repeated online gaming experience might contribute to the altered structure over the amygdala. The decrease of the amygdala gray matter density is often accompanied with individual depression and anxiety, and can lead to an increase of emotion perception, making individuals unable to eﬀectively control themselves. The impairment of amygdala may be associated with individual emotional control and management. The OFC has also been emphasized to be an important area in emotion processing system, it participates in the emotion regulation process of amygdala. The abnormal GMD of this area is mostly found in patients with major depression. A combined functional and structural MRI study [11] provides evidence that the orbitofrontal cortex is a key area in major depression and that structural changes result in functional alterations within the emotional circuit. Prior research [6] has suggested that IAD is more likely to co-occur with other mental health disorder like depression, the decrease of GMD in orbitofrontal cortex, a key area for major depression with major connectivity to other areas of the mood regulation network, may lead to the disability of emotion regulation of IAD.

400

J. Yang et al.

Long-term exposure to online games or watching video will change the brain’s structure and function related to vision and auditory [11]. In this study, it was found that the GMD of the middle occipital gyrus (MOG) was signiﬁcantly reduced in IAD. Han [3] found that both the IAD and professional gamers experienced a signiﬁcant decrease in GMD in the occipital lobe primary visual cortex, which was considered to be caused by individuals’ long-term exposure to visual stimuli such as online games. Our study of internet addiction also found structural abnormality in the visual cortex.

5

Conclusion

Taken together, the present study detected abnormal structural deﬁcits in several brain regions in IAD compared with the HC, which is consistent with published brain image studies on IAD. These brain regions generally present weakened characteristics of IAD in two systems: reward system and emotion regulation system, current ﬁndings might provide insight in understanding the biological underpinnings of IAD. Acknowledgment. This work is supported by the National Natural Science Foundation of China (Grant No. 61105117, 61472058, 61772102).

References 1. Akkermans, S.E., Luijten, M., van Rooij, D., Franken, I.H., Buitelaar, J.K.: Putamen functional connectivity during inhibitory control in smokers and non-smokers. Addict. Biol. 23(1), 359–368 (2018) 2. Fayazi, M., Hasani, J.: Structural relations between brain-behavioral systems, social anxiety, depression and internet addiction: with regard to revised reinforcement sensitivity theory (r-RST). Comput. Hum. Behav. 72, 441–448 (2017) 3. Han, D.H., Lyoo, I.K., Renshaw, P.F.: Diﬀerential regional gray matter volumes in patients with on-line game addiction and professional gamers. J. Psychiatr. Res. 46(4), 507–515 (2012) 4. Hong, S.B., Harrison, B.J., Dandash, O., Choi, E.J., Kim, S.C., Kim, H.H., Shim, D.H., Kim, C.D., Kim, J.W., Yi, S.H.: A selective involvement of putamen functional connectivity in youth with internet gaming disorder. Brain Res. 1602, 85–95 (2015) 5. Jin, C., Zhang, T., Cai, C., Bi, Y., Li, Y., Yu, D., Zhang, M., Yuan, K.: Abnormal prefrontal cortex resting state functional connectivity and severity of internet gaming disorder. Brain Imaging Behav. 10(3), 719–729 (2016) 6. Kim, Y.J., Jang, H.M., Lee, Y., Lee, D., Kim, D.J.: Eﬀects of internet and smartphone addictions on depression and anxiety based on propensity score matching analysis. Int. J. Environ. Res. Public Health 15(5), 859 (2018) 7. Ko, C.H., Hsieh, T.J., Wang, P.W., Lin, W.C., Yen, C.F., Chen, C.S., Yen, J.Y.: Altered gray matter density and disrupted functional connectivity of the amygdala in adults with internet gaming disorder. Prog. Neuro-Psychopharmacol. Biol. Psychiatr. 57, 185–192 (2015)

Brain Structural Abnormalities in Reward and Emotion System in IAD

401

8. Lin, X., Dong, G., Wang, Q., Du, X.: Abnormal gray matter and white matter volume in internet gaming addicts. Addict. Behav. 40, 137–143 (2015) 9. Pan, N., Yang, Y., Du, X., Qi, X., Du, G., Zhang, Y., Li, X., Zhang, Q.: Brain structures associated with internet addiction tendency in adolescent online game players. Front. Psychiatr. 9, 67 (2018) 10. Qi, Q., Wang, W., Deng, Z., Weng, W., Feng, S., Li, D., Wu, Z., Liu, H.: Gray matter volume abnormalities in the reward system in ﬁrst-episode patients with major depressive disorder. In: International Conference on Advanced Machine Learning Technologies and Applications, pp. 704–714. Springer (2018) 11. Scheuerecker, J., Meisenzahl, E.M., Koutsouleris, N., Roesner, M., Sch¨ opf, V., Linn, J., Wiesmann, M., Br¨ uckmann, H., M¨ oller, H.J., Frodl, T.: Orbitofrontal volume reductions during emotion recognition in patients with major depression. J. Psychiatr. Neurosci. (JPN) 35(5), 311 (2010) 12. Tang, W., Zhu, Q., Gong, X., Zhu, C., Wang, Y., Chen, S.: Cortico-striato-thalamocortical circuit abnormalities in obsessive-compulsive disorder: a voxel-based morphometric and fMRI study of the whole brain. Behav. Brain Res. 313, 17–22 (2016) 13. Wang, H., Jin, C., Yuan, K., Shakir, T.M., Mao, C., Niu, X., Niu, C., Guo, L., Zhang, M.: The alteration of gray matter volume and cognitive control in adolescents with internet gaming disorder. Front. Behav. Neurosci. 9, 64 (2015) 14. Weinstein, A., Livny, A., Weizman, A.: New developments in brain research of internet and gaming disorder. Neurosci. Biobehav. Rev. 75, 314–330 (2017) 15. Weng, C.B., Qian, R.B., Fu, X.M., Lin, B., Han, X.P., Niu, C.S., Wang, Y.H.: Gray matter and white matter abnormalities in online game addiction. Eur. J. Radiol. 82(8), 1308–1312 (2013) 16. Yuan, K., Qin, W., Wang, G., Zeng, F., Zhao, L., Yang, X., Liu, P., Liu, J., Sun, J., von Deneen, K.M., et al.: Microstructure abnormalities in adolescents with internet addiction disorder. PLoS ONE 6(6), e20708 (2011) 17. Zhou, Y., Lin, F.c., Du, Y.s., Zhao, Z.m., Xu, J.R., Lei, H., et al.: Gray matter abnormalities in internet addiction: a voxel-based morphometry study. Eur. J. Radiol. 79(1), 92–95 (2011)

Research on Optimization of Rescue Resource Scheduling in Inter-regional Integrated Transportation Network Under Emergency Xiaoqian Peng1, Xueting Li1,2,3(&), Guobao Du4, and Jingchun Geng5 1

3

School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China [email protected] 2 National Railway Train Diagram Research and Training Center, Southwest Jiaotong University, Chengdu 610031, China National and Local Joint Engineering Laboratory of Comprehensive Intelligent Transportation, Southwest Jiaotong University, Chengdu 610031, China 4 Transport Ofﬁce, China Railway Chengdu Bureau Group Co., Ltd., Chengdu 610031, China 5 China Railway Design Corporation, Tianjin, China

Abstract. Emergency resource scheduling is the key to emergency rescue system. Fast, accurate and effective rescue resource scheduling is one of the important links in disaster relief and disaster reduction, which directly affects the effectiveness of disaster relief and disaster reduction. Under the condition of multiple emergency points and transportation capacity limitations after the occurrence of emergencies, the emergency rescue scheduling optimization model under emergencies is established, and the hierarchical sequence method is designed to solve the model. Keywords: Emergency rescue

Resources Resource scheduling

1 Introduction Emergency resource dispatching is a key link in emergency rescue, which runs through the beginning and end of emergency rescue. The research on emergency resource scheduling is to carry out emergency resource dispatching quickly, accurately and effectively, and to complete the disaster relief and disaster reduction work faster and better, so as to eliminate or reduce the loss of the disaster system as much as possible, and avoid untimely and unreasonable scheduling. The cause caused the event to deteriorate, which led to other problems. Many experts and scholars have studied emergency dispatching problems from different angles and achieved certain results [1–3]. Literature [1] studied the process of emergency resource scheduling problem with multiple emergency supply points as single demand points. According to the target difference of emergency resource dispatching, the literature [2] divides transportation into two parts: front-end transportation and back-end transportation. Literature [3] proposed a scheduling scheme for coordinated optimization of multi-cycle emergency © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 402–407, 2019. https://doi.org/10.1007/978-3-030-04582-1_47

Research on Optimization of Rescue Resource Scheduling

403

resources in multiple epidemic areas. Judging from the existing research results, the emergency rescue dispatch research of emergency has not considered how to dispatch the right amount of resources under the condition of limited transportation capacity. In view of the deﬁciencies of the existing research, the following will consider the transport capacity to establish a resource scheduling model that meets resource requirements.

2 Problem Description The actual operation of emergency resource dispatch is to determine the emergency point of participation in the rescue and the amount of resources and transportation time for transportation during emergency rescue [4]. When establishing the emergency resource scheduling model, starting from the characteristics of emergency resource dispatching, we must ﬁrst consider the urgency of time [5]. Therefore, in the scheduling model of this paper, the shortest scheduling time is the target; at the same time, in the process of scheduling, the emergency resource transportation capacity is a constraint condition, that is, the capacity of onetime transportation is limited, and the number of times of scheduling resources from the same resource supply point may be more than one time, which complicates the scheduling model, but is more consistent with the actual scheduling process. On this basis, consider scheduling emergency resources during the restricted period. Under normal circumstances, the most basic scheduling idea is to adopt the principle of proximity, that is, after the occurrence of an emergency, ﬁnd the resource supply point closest to the emergency, and then select the resources required by the affected area for scheduling. However, this approach is only feasible when the resources available from the nearest resource point in the affected area can meet the needs of the affected area. However, for large-scale emergencies, this assumption is usually not established, because the emergency resources required for emergencies are often large, and the resource storage space of resource supply points is limited, and the resource storage capacity is difﬁcult to meet the needs of the affected areas. Therefore, only one resource supply point often cannot meet the demand for emergency resources in the affected area. Therefore, multiple emergency resource supply points should be considered for emergency rescue [6]. Because the emergency response time of the emergency is relatively high, the shortest emergency dispatch time is selected as the target. In this process, you need to choose an appropriate time limit, that is, specify a time, the emergency dispatch time exceeds this time, because the rescue plan under the condition of exceeding the limit period has little rescue effect, and the satisfaction is almost zero, and this time is the limit. Period, the emergency resource scheduling within the restricted period is valid, otherwise it needs to be excluded. This paper discusses the emergency resource scheduling under the condition of limited period, and takes the shortest time as the objective function. In most practical situations, due to transportation capacity limitations, transportation vehicles cannot transport emergency resources needed in the affected area to their destinations at one time. Therefore, for the transportation of the amount of resources stored in a resource supply point, the transportation of the resources cannot be

404

X. Peng et al.

completed at one time, and multiple round-trip transportation is required. A simple emergency resource scheduling model cannot satisfy this situation. Therefore, this paper gives an emergency resource scheduling model with transport capacity constraints during the restricted period, and considers that the same emergency resource supply point may participate in multiple dispatches, which is closer to the actual situation.

3 Establishment of Optimization Model and Solution Ideas In order to facilitate the research, the following assumptions are made on the resource scheduling problem under the condition of transportation capacity limitation after the occurrence of an emergency: Hypothesis 1: Resources with more than one resource supply point may be transported more than once. This assumption can reflect resource transport capacity constraints. Hypothesis 2: After an emergency occurs, the transportation path from the resource supply point to the emergency point is determined. Hypothesis 3: The same resource supply point transport different resources to the same emergency point for the same time. Hypothesis 4: It is assumed that the type and amount of emergency resources required for each emergency point at the time of scheduling is known [7]. Hypothesis 5: Assume that, under the condition of meeting the restriction period, in addition to satisfying the total resource storage of the resource storage point is greater than or equal to the total demand of the emergency point, and the actual transportation volume of each resource supply point must be equal to the emergency resource demand, The plan is feasible. This will ensure that resources are not wasted. 0 zik ; zik xik zik ¼ ði ¼ 1; 2; ; n; k ¼ 1; 2; ; sÞ; which can be constructed xik ; zik [ xik from the above assumptions Optimization model: 0

minT ð/Þ ¼ tij

ð1Þ

Research on Optimization of Rescue Resource Scheduling

405

Variables and parameters of the model are deﬁned as follows: Aj : emergency point that needs rescue, (j = 1, 2, …, m). Bi : emergency resources store point, (i = 1, 2, …, n). sj : emergency resource types required for emergency point Aj , (k = 1, 2, …, s; j = 1, 2, …, m). Xkj : demand of emergency point Aj for the resources kj , (i = 1, 2, …, n; k = 1, 2, …, s). yi : 0–1 variable, take 1 when the resource supply point Bi participates in the rescue, otherwise take 0. zik : The maximum amount of one scheduling when scheduling the resources kj from the resource supply point Bi , (i = 1, 2,…, n; k = 1, 2,…, s). tij : Transportation time of resources from resource storage point Bi to Emergency point Aj . 0 tij : The total time from the resource supply point Bi to the emergency point Aj to be 0

dispatched, (tij is an integer multiple of tij ). Tj : The deadline for the dispatch of the emergency point Aj speciﬁed in the case of 0 an emergency, (0< tij Tj ). Equation (1) is the objective function of emergency resource scheduling, which represents the scheme with the shortest scheduling time among all feasible scheduling schemes, that is, the optimal scheme. formula (2) ensures that the number of resources k that all resource supply points can provide is greater than or equal to the number of resources k required for all emergency points, that is, the requirements of all emergency points can be satisﬁed. formula (3) guarantees the scheduling of resource supply points participating in emergency rescue under the conditions of transportation capacity constraints. The total amount of the resource k is equal to the amount of the resource k required in the disaster area. the formula (4) makes the longest dispatch time of all the emergency resource supply points Bi participating in the rescue to the disaster point Aj not exceed the limit period. Equation (5) represents the 0–1 variable. When yi = 1, it indicates that the resource supply point Bi participates in the scheduling of some emergency resources. When yi = 0, it indicates that the resource supply point Bi does not participate in the rescue. This section studies the multi-resource scheduling of multiple resource supply points participating in multiple emergency points. To make the results clearer, any scheduling scheme / can be expressed in the form of vectors and matrices as: .. . / ¼ x1; x2; . . .; xmT; xj ¼ Xj110 Xj120 Xj1s0 .. . ...Xjn10 Xjn20 Xjns0 ; 0 Xik 0 Xik xj represents the scheduling scheme of the emergency point Aj , Xjik represents the maximum resource amount of the resource supply point Bi in the restriction period Tj to 0 schedule the resource sj to the emergency point Aj , and Xjik indicates the resource supply when the scheme / is adopted. The point Bi schedules the actual scheduling amount of the sj resource to the emergency point Aj . It can be concluded from the scheme xj that the resource supply point Bi participating in the scheduling schedules the actual number of the resources sj to the

406

X. Peng et al.

emergency point Aj , and the row i indicates the number of various resource schedules when the resource supply point Bi participates in the emergency point Aj scheduling, column s indicates the number of resource kj when different resource supply points 0 participate in the scheduling of the emergency point Aj . (If Xjik = 0, it means that the resource k of Bi is not dispatched to the emergency point Aj ).

4 Algorithm Design When solving the model, we can ﬁrst consider the scheduling of a resource supply point. The algorithm is as follows: If only one resource of the supply point B1 meets the requirements of the emergency point, that is, t1 T and x1 X is satisﬁed. ( x1 x1 h i z1hi1; z1 2 Z x1 x1 n1 = , x1 x1 z1 represents a rounding operation on z1 . Therefore, it can 2 6 Z ; z1 z1 be divided into two cases: if t1 þ 2n1 t1 T, there are / = fðA1 ; X Þg and Tð/ Þ ¼ ð2n1 þ 1Þt1 ; if t1 þ 2n1 t1 [ T, there is no optimal solution. The above is a solution to a resource supply point, the process is fairly simple, but this paper studies the scheduling problem of multiple emergency resource supply points. If you use the above method to solve the problem, the problem will become quite complicated, so this paper chooses another algorithm, the algorithm is described as follows: Assume that the maximum number of times the k supply point is h resource i scheduled is nk . Let 2nk tk tk ¼ T and solve for nk =

T þ tk 2tk

. If nk zk xk , the number

of times of the resource k supply point is nk . If nk zk [ xk , the number of times of the ( xk xk zkhi1; zk 2 Z 0 resource k supply point is nk = . The transport time Tk corresponding to xk xk 2 6 Z ; zk zk Bk is tk ; 3tk ; . . .; ð2nk 1Þtk ; if nk zk xk , the corresponding transfer amount mk is zk ; zk ; . . .; zk ; When nk zk [ xk , the corresponding amount of each adjustment mk is zk ; zk ; . . .; xk ðnk 1Þzk , (i = 1, 2, …, nk ). 0 Tij is the scheduling i for the resource supply point j, the earliest time when the resource reaches the emergency point. mij is the amount of resource transportation for the resource supply point j when the scheduling i is performed. Each shipment has a 0 corresponding Tij and mij . Step 1: Sort the s resources that need to be scheduled according to the urgency of resource demand. The urgency is high and the scheduling is ﬁrst. Step 2: For the scheduling time of different scheduling methods of the resource k, 0 0 0 arrange Tij in the order of small to large, if there is Tij = Ti0 j0 , turn to Step three. 0

0

0

0

Step 3: If mij [ mi0 j0 , Tij is placed before Ti0 j0 ; if mij \mi0 j0 , Tij is placed after Ti0 j0 , if mij ¼ mi0 j0 , turn to step four. 0 0 0 0 0 Step 4: Tij = Ti0 j0 and mij ¼ mi0 j0 , if i [ i0 , Tij is placed before Ti0 j0 , otherwise Tij is 0

placed after Ti0 j0 , if i = i0 , turn to step ﬁve (j 6¼ j0 ).

Research on Optimization of Rescue Resource Scheduling

407

Step 5: Select any one of the schemes for scheduling. Step 6: End. k k P P The above order is sorted for the number of s = ni , and the above s = ni i¼1

i¼1

numbers are rearranged, then there are P1 ; P2 ; . . .. . .; Ps ; Then the number Pi of the position i is the corresponding trafﬁc volume, and then mi j can be determined 0 according to the subscript of Ti j ranked at the position i, at this time, Pi ¼ mi j

5 Conclusion Emergency rescue is an extremely complicated research system. Based on the existing research, the article considers the characteristics of emergency dispatch and the urgency of scheduling time. The hierarchical sequence method is designed to solve the model. The problem of multiple emergency point multi-resource scheduling is very complicated. This paper only studies one of the cases. For other scheduling methods, such as the scheduling of equipment with continuous rescue capability, further research is needed. Acknowledges. This research was supported by the National Key R&D Program of China (2016YFC0802208), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), and Science and Technology Plan of China Railway Corporation (Project No.: 2016X006-D), and Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017-RK00-00378-ZF).

References 1. Liu, C., Shi, J., He, J.: Study on optimization model of a class of emergency material scheduling. China Manag. Sci. 9(3), 29–36 (2001) 2. Song, M.: Construction of emergency disaster relief logistics distribution system model. Taiwan: Master’s thesis of Guoli Jiaotong University (2005) 3. Wang, X., Wang, H.: Multi-cycle emergency materials collaborative optimization scheduling in multiple epidemic areas. Syst. Eng. Theory Pract. 32(2), 283–291 (2012) 4. Liu, L.: Research on issues related to enterprises participating in emergency management of natural disasters. Yanshan University, Qinhuangdao (2009) 5. Chen, C.: Research on emergency resource scheduling model and algorithm for emergency. Nanjing University of Aeronautics and Astronautics, Nanjing (2010) 6. Wang, Z.: Emergency material allocation model based on continuous soft time window restriction. Huazhong University of Science and Technology, Wuhan (2011) 7. Lin, W.: Research on highway emergency rescue resource dispatching system based on electronic map. Southeast University, Nanjing (2010)

Vehicular Applications and Others

The Applied Analysis of Big Data in Trafﬁc Safety Management Xiuzhen Yu(&), Ruifang Mu, Rui Yang, and Lieni Wang School of Trafﬁc and Logistics, Southwest Jiaotong University, 999 Xi’an road, Chengdu 611756, China [email protected], [email protected], [email protected], [email protected]

Abstract. With the development of the Internet and trafﬁc information technology, more and more trafﬁc data resources are obtained, and the scale is increasing, transportation industry has entered the era of big data. How to dig out effective information from massive data to provide services for trafﬁc congestion and security problems has been paid more and more attention. First, this paper analyses the source, content of trafﬁc safety data, and the general process of data mining; Secondly, the application of big data in the ﬁeld of trafﬁc safety is discussed; Finally, an application framework of trafﬁc safety big data based on data processing is put forward. Keywords: Big data Data mining Application framework

Trafﬁc safety analysis

1 Introduction With the rapid development of our economy and the acceleration of urbanization process, the number of motor vehicles has increased dramatically, followed by problems such as trafﬁc jam, frequent trafﬁc accidents. It has been unable to solve the problem of trafﬁc management by manpower. On the other hand, trafﬁc sensor and information equipment such as GPS, IC card, cameras has spread the whole city, these devices has collected vast amounts of data for the trafﬁc managers every day. More and more trafﬁc managers begin to think about applications of big data. Big data technology not only has a certain advantage in data acquisition, storage and processing, but also offers platform for data integration, information integration, application integration. The role of big data in trafﬁc safety management is mainly embodied in hazard identiﬁcation, safety data analysis. Hazard identiﬁcation focuses on prior recognition, the active hazard identiﬁcation system developed by Beijing MTR Corporation is a good example; in terms of safety data, it summarizes and forecasts the trafﬁc safety situation through summary analysis of the data collected by equipment and case analysis of counterparts at home and abroad. The source of big data and data mining process are analyzed in this paper, and the application of big data analysis in trafﬁc safety analysis and evaluation are discussed in detail, put forward the application architecture of trafﬁc big data based on data mining © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 411–419, 2019. https://doi.org/10.1007/978-3-030-04582-1_48

412

X. Yu et al.

process, which will provide the reference for further research as the big data applied to the management of trafﬁc safety.

2 Trafﬁc Safety Data Analysis 2.1

The Source and Content of Trafﬁc Safety Data

Trafﬁc safety data is the basis for describing and analyzing trafﬁc safety conditions, including accident data, environmental data, trafﬁc violations data, trafﬁc flow data, trafﬁc infrastructure data, and so on, among them accident data is the basis of safety analysis. Trafﬁc safety data is classiﬁed in accordance with access and produce elements, its classiﬁcation and contents are shown in Table 1 [1, 2].

Table 1. Trafﬁc safety data sources and content. No. Source 1 According to provider

2

According to the access methods

Classiﬁcation public security department transport Sector

Information content motor vehicle, driver, trafﬁc violation, trafﬁc accident, control facilities, police car, etc. volume of trafﬁc, information related to charges, road condition information, implementation information, transport management information, Trafﬁc control information, etc. operating company network information, road segment information, transport infrastructure information, video information, information related to charges, etc. the information such as meteorological meteorological department monitoring, prediction, the rain, snow and ice, fog public information etc. road condition, accident information etc. Other rescue resources information, etc. collected automatically recorded video by monitoring probes, the images collected by automatic identiﬁcation by monitoring equipment of license plate in all kinds of equipment bayonet, all kinds of images captured by electronic police equipment, trafﬁc flow collected by sensor, GPS trajectory, automatic payment information, etc. artiﬁcial collection the vehicle and the driver’s information collected by law enforcement ofﬁcer, the illegal information investigated and punished obtained by road planning, all kinds of vehicles and administrative means drivers’ archives information, etc.

The Applied Analysis of Big Data in Trafﬁc Safety Management

413

Authorities such as the ministry of public security in our country, the national standardization committee make a series of standards and rules for acquisition projects, storage form of data such as trafﬁc accident, etc. [2]. 2.2

The Data Mining

Data mining is the process that extracts the information or pattern implicit, not been informed in advance but potentially useful from a lot of practical application data, which is incomplete, ambiguous, and of great randomness [3]. We can ﬁnd knowledge that users might be interested in by big data mining in a large number of chaotic data. Due to trafﬁc safety data from different departments have their own metrics, this leads to different data structure type, such as geography, images, video, text, etc. As a result, the primary task to realize the process is the preprocessing of diversity of data, which adopts the uniﬁed standard speciﬁcation or codes data information to realize information sharing across different departments and regions, and then data mining algorithm is used to identify the factors influencing the trafﬁc safety. Data mining technology uses proper data mining methods to analyze the close degree of link among large amounts of data on the premise of complete and normative data, and then knowledge reasonable is given [4]. Acquired knowledge will be presented to the users in the form that understood, used and viewed easily. Data mining is a process of decision support, it mainly bases on technology such as artiﬁcial intelligence, machine learning, pattern recognition, statistics, database and visualization, etc. It can analyze autonomously the data to make inductive reasoning, and then dig out the potential and valuable information, which will provide intellectual support for the user to make a decision. According to the data flow, data processing can be divided into three steps, as shown in Fig. 1.

Knowledge

Data input

data preprocess Data integrated

Data fusion calculation

data mining

generativ e rule

asocciation rules genetic algorithm artificial neural network decision-making

ruler expressions visualization data query

data conversi data correlation situational database

Fig. 1. Data processing

output

414

X. Yu et al.

3 The Application of Big Data in the Trafﬁc Safety Management With the trafﬁc safety data accumulation and continuous improvement of the information system, the traditional analysis method cannot meet the needs of the user analysis, we need to use the more advanced data analysis method to analyze and process the data, and dig out the deep cause of the accident. 3.1

The Application of Big Data in the Analysis of Distribution of the Trafﬁc Accident and the Accident Risk Study and Judge

The analysis of trafﬁc accident distribution and risk assessments needs to deal with a lot of historical data and the corresponding trafﬁc flow parameters [5]. Common data acquisition technology and computer network technology cannot meet the requirements. Big data technology provides the technical support for the distribution and forecast of the trafﬁc accident to ensure that the processing, analysis and storage of relevant data are carried out at a high speed. (1) The analysis of trafﬁc accident distribution based on big data By extracting the accident information such as the time of accident, the accident site from trafﬁc accident database, the time and regional distribution of the historical accident can be got. According to the distribution, we can forecast the accident in the future by the adoption of a certain mathematical model. The distribution of accident varies by the different databases established. For example, if a year is divided into holidays and working days or one day is divided into trafﬁc peak and off-peak interval, we will get accident distribution during the holidays, working days, and the distribution of accident during peak and off-peak hours, etc. According to the location of accidents, we will get accident black point, road and area. Based on the analysis and prediction of trafﬁc accidents distribution, trafﬁc management departments may focus on screening for accident black belt, and the reasonable reconstruction of trafﬁc facilities (such as trafﬁc lights, marker, signs, etc.), strengthen police force to guide and direct trafﬁc flexibly during the peak time of the accident. (2) The study and judge of trafﬁc accident risk based on big data Trafﬁc flow status has a close connection with trafﬁc safety. Studies have shown that the parameters of trafﬁc flow state such as trafﬁc volume, speed and large vehicles proportion will affect trafﬁc safety [6]. Multi-source heterogeneous trafﬁc flow data processing by extracting the trafﬁc flow data from monitoring station before a certain type of accident and normal data was integrated to get characteristic variable of this type of accident. Using characteristic variable sets as input variables, a real-time accident risk prediction model is established to predict the possibility of this type of accident at the next moment. If there is possibility of an accident next moment, we can use changeable message sign, trafﬁc information broadcasting system, vehicle navigation system, text message to remind drivers to drive carefully.

The Applied Analysis of Big Data in Trafﬁc Safety Management

3.2

415

The Application of Big Data in the Cause Analysis of Trafﬁc Accident

Trafﬁc accident causes roughly includes four factors of person, vehicle, road and environment, each factor also can be subdivided into several layers, the lower level the more detailed description. Speciﬁc performance is shown in Table 2. Table 2. Causes of trafﬁc accident Human Negligence

Inattention; Lags in response; Error judgement

Road

Pavement type; Intersection; Pavement status Landform Road grade Environment Trafﬁc Speed Driver illegal Violating trafﬁc environment Trafﬁc overtaking; regulations and flow Overspeed and other Trafﬁc management rules overweight; sign and Driving without license; marking Drive when intoxicated Trafﬁc control conditions Natural environment Ride man Cycling in the fast lane; Pitchout; Pedestrian Vehicle Lack of maintenance jaywalk; and inspection Across the separation Fall into disrepair barrier improper adjustment Response error Improper operation Faulty movement

Using big data technology to analyze the causes of trafﬁc accident has the following advantages: First, through the analysis of the causes of trafﬁc accidents, we can know which factors have a greater impact on trafﬁc accidents and which factors will appear together in a certain type of accidents. For example, the change of vehicle running state is one of the main factors causing accidents, which mainly divided into: going straight, turning left, turning right, parking, reversing, turning round, changing lanes, etc. Through the analysis of the big data, it can be found that accident probability is the greatest when a vehicle turns left, followed by turning right, then is changing lanes and turning around, and ﬁnally the astern and other conditions. Grasping the effect degree of different driving condition to the accident, trafﬁc administration can weigh weight accordingly, adjust the time and location of trafﬁc lights at intersections more reasonably, and plan trafﬁc markings and vehicle cornering guide lines, etc. [7].

416

X. Yu et al.

Second, adopting the association rules method in data mining can get the relationship between different factors, such as the relationship between the intersection type and the accident vehicle types in the above analysis. Analysis shows that intersections with a small number of bifurcation often lead to large vehicle accident because of the narrow ﬁeld of vision, small area of the road, intersecting flows of pedestrians and vehicles, etc. Therefore, for large vehicles management, it can be guided and propagandized to avoid them to enter into a small intersection [8]. Third, through the data mining method, identifying the key elements leading to accidents from the basic data collected can provide selective opinions of index for trafﬁc safety evaluation model. 3.3

The Application of Big Data in the Traveling Safety Assessment

Along with the development of intelligent transportation system, there are many kinds of trafﬁc sensors at present. they store a lot of meteorological data under the bad weather situation trafﬁc flow data of bad trafﬁc flow state in the long-term work, based on the data processing the possibility of accident can be calculated because of the bad weather and brake not timely under the condition of car-following operation by using parking stadia model, thus provide technical decision support for formulating a reasonable vehicle spacing and speed. For example, guide the driver to travel smooth at a reasonable speed by taking a certain standard of speed limit, especially on the road of the downhill, corner, etc. and annunciating the speed limits through roadside complete equipment such as trafﬁc signs, electronic information card, reduce the speed difference in the different sections, ensure the safety of driving. 3.4

The Application of Big Data in the Evaluation of Road Network Safety Performance

Vast amounts of trafﬁc safety data is the basis of trafﬁc safety analysis, at present the United States department of transportation has counted up and built the trafﬁc safety performance SPFs function based on trafﬁc safety data from the state department of transportation, and used the SPFs function to evaluate the trafﬁc safety performance of states according to the influence on trafﬁc safety caused by linear combination in the highway system, weather factors, driver behavior changes, and then carry out targeted technology research of trafﬁc safety performance boost. The purpose of the regional road network safety assessment is to investigate regional road network trafﬁc safety. After regional road network being built and running, the factor of main construction consisting of pavement, road line is relatively stable, but the operation of the trafﬁc flow, weather conditions, trafﬁc participants are the main factors influencing the safety of the regional road network. Trafﬁc flow parameters, including road average speed, speed difference and lane occupancy ratio can be obtained through the ﬁeld collection devices, security parameters such as trafﬁc accident statistical data, accident rate data can be obtained from the trafﬁc police department, visibility, amount of precipitation, snowfall and other meteorological parameters can be obtained from the meteorological department, on the basis of these large amounts of data the evaluation indexes can be quantiﬁed, so as to reach the purpose of quantitative evaluation.

The Applied Analysis of Big Data in Trafﬁc Safety Management

417

4 The Framework of Big Data Application System for Trafﬁc Safety Management The efﬁciency and safety of modern transportation must rely on the strong support of the informational and digital platform. Trafﬁc safety management based on big data mainly relies on “cloud computing” information technology, and makes prospective analysis and forecast of urban trafﬁc safety at the more macroscopic, comprehensive and three-dimensional angle. On the basis of the comprehensive consideration of the characteristics and application of trafﬁc data, this paper designs a framework of big data application system for trafﬁc safety, as shown in Fig. 2. The framework is based on the big data processing process of trafﬁc safety shown in Fig. 1, including four layers, which are data source layer, data processing layer, knowledge layer and application layer, respectively.

Data source layer

Traffic basic data

Traffic investigated data

Traffic accident data

Road detection Data

GPS

ETC data

RFID Video

MapReduce Off-line Analysis data mining

HDFS

Results show

Traffic model prediction Data processing layer

Sqoop data integration

RelaƟonal database

Hbase database

Storm Online processing Flow analysis

Results show

Traffic flow prediction

Knowledge layer

application layer

Road network structure

Traffic anomaly detection

Vehicle running rules

Traffic safety early warning

Daily travel behavior law

Study on the risk of accident

Traffic accident law

Driving safety assessment

...

Road network safety performance evaluation

Fig. 2. Framework of big data application system for trafﬁc safety management

(1) Data source layer. Trafﬁc heterogeneous data include structured and unstructured data [9]. These data sources mainly include radio frequency identiﬁcation (RFID), video surveillance, GPS trajectory data, trafﬁc application service data and so on. (2) Data processing layer. Data processing layer is the core of the big data processing system framework. Sqoop integrates source data and stores it in distributed database HBase. The storage system of HBase and Impala based on HDFS is used. Impala provides real-time interactive SQL query function, and queries data

418

X. Yu et al.

directly from HBase with SELECT, JOIN and statistical functions to achieve rapid big data storage and analysis [10, 11]. According to different trafﬁc requirements, two sets of computing frameworks can be used. The MapReduce off-line computing framework is used for prediction of trafﬁc model and mining of trafﬁc operation rules. The Storm real-time trafﬁc flow calculation framework is used to deal with real-time trafﬁc flow data. (3) Knowledge layer. Processing trafﬁc large data is to analyze the data deeply and dig deep knowledge contained in the data to ﬁnd the law hiding internal data, which includes the daily travel behavior law, road network structure, trafﬁc accident law, vehicle running rules and so on. For video and image data, CUDA architecture is adopted to extract its features and abstracts quickly for mining analysis. The data of electronic charge, application service and GPS are analyzed and excavated in parallel with Mahout based on MapReduce computing model, then knowledge is formed for higher level of trafﬁc Security management. (4) Application layer. All the research will be attributed to the application. Barge data technology has played a great role in promoting the further development of intelligent trafﬁc safety, such as trafﬁc safety early warning, accident prediction, trafﬁc anomaly detection, road network safety performance evaluation and so on.

5 Conclusion The technology of big data and its application can provide support for formulating a reasonable scientiﬁc decision on trafﬁc safety. But the current research focuses on theoretical research, and how to apply theoretical ﬁndings to practical safety management is the goal of the next stage. In general, the introduction of the concepts and methods of big data will improve the ability to solve the problem of road trafﬁc safety further, but at the same time it puts forward higher requirements on trafﬁc safety managers. Acknowledgement. This work was supported by National Key R&D Program of China (Grant No. 2016YFC0802209).

References 1. Ding, H.J.: The Practical Inquiry of Big Data in The Public Security Trafﬁc Management, in the 10th academic conference of social sciences circles in Tianjin with the topic of scientiﬁc development, collaborative innovation, dream building 2014: Tianjin, China. 6 2. Zhao, X.Y.: The Assessment Methods of Highway Trafﬁc Safety Based on The Multi-source Heterogeneous Data, 2013, Harbin Institute of Technology (2013) 3. Jiang, S.Y., Li. X., Zheng, Q.: Data Mining Theory and Practice, pp. 3–5. Electronic Industry Press, Beijing (2011) 4. Valtehev, P., Missaoui, R., Godin, R.: Formal concept analysis for knowledge discovery and data mining: the new challenges. In: Eklund, P. (ed.) ICFCA, pp. 352–371. Springer, Heidelberg (2004)

The Applied Analysis of Big Data in Trafﬁc Safety Management

419

5. Liu, M.Z.: Trafﬁc Safety Analysis Based on Data Mining Technology, Shanghai Jiaotong University (2014) 6. Hu, C.Y., Yang, X.M.: The information service of comprehensive transport hub based on big data. Comprehensive Transportation, 2015 (7), 60–62 + 70 7. Zhang, H., etc.: The system architecture of intelligent transportation based on Big Data. J. Lanzhou Univ. Technol. 41(2), 115–112 (2015) 8. Shi, Q., Abdel-Aty, M.: Big data applications in real-time trafﬁc operation and safety monitoring and improvement on urban expressways. Transp. Res. Part C 58, 380–394 (2015) 9. Ouyang, Q.M., Wu, C., Huang, L.: Methodologies, principles and prospects of applying big data in safety science research. Saf. Sci. 101, 60–71 (2018) 10. Battiato, S., Farinella, G.M., Gallo, G., Giudice, O.: On-board monitoring system for road trafﬁc safety analysis. Comput. Industry 98, 208–217 (2018) 11. Fatholahzade, N., Akbarizadeh, G., Romoozi, M.: Implementation of Random Forest Algorithm in Order to Use Big Data to Improve Real-Time Trafﬁc Monitoring and Safety. J. Adv. Comput. Eng. Technol. 4(2), 1–10 (2018)

An Innovative Design and Simulation of Transom Type Venturi Cooling Design for High-Power LED Headlamp Maw-Tyan Sheen(&) and Qian-ting Wang School of Mechanical and Automotive Engineering, Fujian University of Technology, The 3rd Street Minhou County, Fujian Academy Road, Fuzhou, China [email protected]

Abstract. This research an innovative window type Venturi cooling system to help reduce the high power LED emitting heat on working. It uses lamp-shape heat sink base that is coated by aluminum nitride (AlN) ceramics to be an insulation layer which can reduce the damage of emanating heat and promote the working hours of high power LED. It is obviously seen that the force convection contributes a lot on heat dissipation when vehicles running. The record shows that the highest temperature comes to 96.4 °C, 93.3 °C and 89.8 °C on the high power LED headlamps of 85 W, 80 W and 100 W separately without the convection window design. The other record shows that the temperature comes to 74.6 °C, 73.7 °C and 64.8 °C on the same lamps but with cooling down window embedded. The effective cooling down temperature is 21.6 °C, 19.8 °C and 25 °C. The heat dissipation performance contributes 22.61%, 21% and 27.83%. By the analysis of ﬁnite elements on window type Venturi system shows that experiments data coincides with theory assumption. This project absolutely shows that an innovative window type Venturi cooling system is highly effective on heat dissipation and can provide more competitive high power LED products in the market. Keywords: High power LED headlamp Aluminum nitride (AlN) Venturi cooling device Force convection F.E.M

1 Introduction The main reasons of high power LED headlamp adopted in motorcycles and automobiles are its environmental protection character and long-term usage. But one problem is concerned on the heat dissipation which makes the cost up about high power LED headlamp. The next prospect is to reach feasible and reasonable cost down of design to improve heat performance. This project proposed an innovative window type Venturi design to develop brand new heat dissipation system for high power LED headlamp. This article provides technical information exchange with others and could offer promoting the reliability and quality for other relative products. There are also theory model construction, simulation analysis and research results included in the project. The fabrication of study here could also promote the technology of high power © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 420–429, 2019. https://doi.org/10.1007/978-3-030-04582-1_49

An Innovative Design and Simulation of Transom Type Venturi Cooling Design

421

LED headlamp and extend the usage of it. Furthermore, the method of project could apply for a patent because of its capability of easily applied in other relative industry ﬁelds. This article is mainly focused on the study of a window type of Venturi using heat sink base coated with aluminum nitride (AlN) ceramics to reduce and dissipate out the heat generated by high power LED through window convection. Before delving into our project, we will retrospect to heat dissipation design and researches on high power LED. In 2004 Narendran and Gu [1] made a lifetime test on LED in different environmental temperatures. In the experiment, they used white light of high power LED of the same fabrication and the result showed that the illuminating efﬁciency decayed exponentially with working time. The main factors lied on temperature and the higher temperature of the surroundings the more intense of decaying becomes. Another experiment showed that different fabrications under the same white light LED and environmental temperature have different impacts on the lifetime of LED. Because different fabrication has different thermal resistance and cause different decaying effects. Based on the report provided above, the fabrication design and working environment temperature have direct impacts on the dissipation performance and lifetime usage on the white light high power LED. In 2004 Park [2] used infrared thermal image instrument to detect chip temperature of LED. He found that the transform efﬁciency reduced 70% as chip temperature rose from 25 °C to 107 °C. This conclusion indicated that handling chip temperature is critical to improve performance of LED. In 2004 Niki [3] also found that the illuminating efﬁciency of LED decreased with working and using times. And overheated junction temperature accelerated the decaying of illuminating efﬁciency. From the above research, we realize the proper heat management is not only necessary on illuminating effect of LED but also crucial for the lifetime of the overall system. In 2005 Shatalove [4] made a thermal analysis on nitride-based deep ultraviolet LED using flip-chip packaged of 280 nm to ﬁnd out bottleneck of heat dissipation. The substrate dimension is 100 mm 100 mm and was coated with aluminum nitride (AlN) ceramics with ﬁn-type heat radiation below. The result showed that the thermal resistance was 3 °C per watt. Through simulation analysis indicated the bottleneck of heat dissipation was the junction area of LED. If the area was small and the heat flux increased extra high. Traditional LED package is not good at emanating heat at all. In 2007 Chen [5] improved the fabrication process by jointing the circuit board with heat dissipation slug. This new way of fabrication reduced the overall materials as well as improved heat performance. Besides, it also increased the power to help better illumination. In 2008 Zhang [6] probe into the thermal conductivity of LED headlamp by using ANSYS software. The parameters included in the test are arrangement of LED, the size and thickness of dissipation area, and using different materials of slug, substrate and heat sink. The result showed that the uniform and ring-like arrangement of LED did not have obvious difference in temperature. By expanding the size and adding up thickness of heat sink came to a temperature difference in case of convection coefﬁcient smaller than 50 W/m2 ∙ K If changing the material of slug and substrate into highly coefﬁcient of thermal conductivity would effectively decrease the junction temperature of LED. In 1992 Metz [7] proposed that setting up a chip surrounded ﬁn-type heat sinks of LED

422

M.-T. Sheen and Q. Wang

could dissipate out the generated heat effectively. In 2000 Hochs [8] designed a series of ﬁn-type heat sinks on the substrate of LED could also passed the heat back and out. In 1995 Azar and Madrone [9] made a test and found out that the higher density of ﬁntype heat sinks could lower down thermal resistance under low flowing of convection but caused resistance up once beyond limitation of flowing. From the explorations above showed that both operation and dissipation way have influences on the illuminating efﬁciency and lifetime usage of LED. This article will further explore by developing a new design of window type Venturi to improve thermal performance. Depending on handling heat sink module through experiment and ﬁnite element theory achieves the goal of decreasing the junction temperature and promoting illumination efﬁciency of LED. Based on the accomplishment of the research will contribute to further design for high power LED headlamp of vehicles.

2 Research Methods (1) Gas window type venturi tube Benny principle introduction The principle of the window-type venturi tube is made using the principle of white Nuri, the formula (1), the fluid particles only by the pressure and their weight and gravity of the impact. If the fluid is flowing in the direction and along the cross section of the streamline, the velocity increase can only be due to the lower fluid pressure on the area. The fluid moves at a higher pressure, and the fluid velocity decreases. Thus, at the highest velocity of the flowing fluid, the pressure is lowest and the lowest velocity occurs at the highest pressure, such as the chimney effect. (2), the fluid will move from the high pressure area to the low pressure area, which can produce natural convection, accelerate the cooling. The pressure of the fluid is small and the pressure is low. LED lights, like forced convection cooling LED lights. Benny principle: 1=2qv2 þ qgh þ p ¼ constant

ð1Þ

Where: v = fluid velocity, G = gravitational acceleration, H = height difference in fluid, P = the pressure on the fluid, q = density of fluid Continuous equation A1 v1 = A2 v2 = constant

ð2Þ

Where: A = cross-sectional area of the chimney, V = fluid flow velocity (2) Gas window type venturi ﬁns lamp holder manufacturing The lamp base and the lamp holder are connected to the heat transfer. The headlamp is connected with aluminum ﬁns as the heat dissipation, and the l headlamp is wrapped in the venturi, and the gas window is in contact with the air. Convection, the aluminum ﬁns for cooling, to achieve cooling function. When the steam, locomotive running when the window-type venturi into a forced convection, better heat dissipation.

An Innovative Design and Simulation of Transom Type Venturi Cooling Design

423

Can be the most effective high-power LED heat dissipation, the window-type Venturi headlamp to stick manufacturing method of production steps, as shown in Fig. 1. An innovative transom type venturi cooling design for high-power LED headlamp, as shown in Fig. 2.

Fig. 1. The window-type Venturi headlamp to stick manufacturing method of production steps

(3) Thermal simulation and analysis for high power LED with ﬁnite element method With the help of ﬁnite element package software MARC and MENTAT, the thermal analysis on the heat sink of the window type Venturi could be done thoroughly. By ﬁnite element simulation of LED lamp holder and substrate coated with aluminum nitride (AlN), the heat dissipation process is able to be seen and data could be recorded. The simulation is on the basis of transient thermal conduction theory which includes analysis of heat conduction and air convection but neglect of heat radiation. The ﬁnite element software provides three-dimensional transient thermal conduction method to analyze natural and forced convection of the lamp holder. The ﬁnite way goes with establishing eight grids of three-dimension for the lamp holder and there are total 9324 elements with 17292 nodes inside the model. In Fig. 3 showed the grids establishment of the simulation for the window type Venturi LED lamp holder. With formula (1) of thermal conduction could decide the temperature distribution for the substrate coated with aluminum nitride (AlN). r ðKrTÞ ¼ C q

@T @t

ð1Þ

Using thermal convection formula below could decide the temperature on the surface of boundary. K

@T ¼ hðT TsÞ @n

ð2Þ

In formula (2), parameters K, C, q, h, Ts are thermal conduction coefﬁcients, speciﬁc heat, density, thermal convection coefﬁcient and room temperature respectively. Table 1 showed the property of using materials. In Fig. 4 showed the simulation of temperature rising up and cooling down for the coated substrate of high power LED on work.

424

M.-T. Sheen and Q. Wang

Fig. 2. An innovative transom type venturi cooling design for high-power LED headlamp High power LED LED substrate coating NAl high Thermal material when the insulation layer

LED substrate coating NAl high Thermal material when the insulation layer Window type venturi tube

Fig. 3. Temperature distribution by ﬁnite element simulation for the substrate coated with AlN of high power LED lighted up Table 1. Material property K (W/moC) Al 202 AlN 104 Cu 385 Air – Room Temperature –

C (J/KgoC) 0.797 0.920 0.385 – –

q (Kg/m3) 2700 1000 8900 – –

h (W/m2oC) – – – 5 –

Ts (oC) – – – – 25

Fig. 4. Temperature rising up and cooling down by F.E.M. simulation for the substrate coated with AlN of high power LED lighted up

An Innovative Design and Simulation of Transom Type Venturi Cooling Design

425

(4) Lumen and temperature measurement for the window type Venturi of high power LED The integrating sphere is an instrument used to measure luminous flux, spectral power, color coordination, color temperature, color rendering, peak wavelength, input power, light effect and waveforms. It can also provide three-dimensional thermal image to measure temperature. a. Temperature measurement of high power LED headlamp of 85 W with the window type Venturi Current: 400 mA, Voltage: 38.89 V, Lumen (LOP): 6484 ml, color temperature (CIE) = 5705K Temperature is measured and recorded from 1 to 60 min after LED is lighted on. The result showed that the highest temperature is 74.6 °C for LED as the heat source. The temperature on the lateral edge of LED is 46.4 °C. The highest temperature for aluminum substrate of high power LED without window setting up in the Venturi is 96.4 °C. showed Fig. 5.

85W

80W

100W

Fig. 5. Temperature measurement for the window type Venturi with high power LED headlamp of respectively 85 W, 80 W, 100 W in serial parallel combination

b. Temperature measurement of high power LED headlamp of 80 W with the window type VenturiCurrent: 400 mA, Voltage: 34.89 V, Lumen (LOP): 6184 ml, color temperature (CIE) = 5680 K Temperature is measured and recorded from 1 to 60 min after LED is lighted on. The result showed that the highest temperature is 73.7 °C for LED as the heat source. The temperature on the lateral edge of LED is 41.8 °C. The highest temperature for aluminum substrate of high power LED without window setting up in the Venturi is 93.3 °C. showed Fig. 5. c. Temperature measurement of high power LED headlamp of 100 W in serial parallel combination with the window type VenturiCurrent: 400 mA, Voltage: 26.01 V, Lumen (LOP): 5462 ml, color temperature (CIE) = 5851K Temperature is measured and recorded from 1 to 60 min after LED is lighted on. The result showed that

426

M.-T. Sheen and Q. Wang

the highest temperature is 64.8 °C for LED as the heat source. The temperature on the lateral edge of LED is 40.3 °C. The highest temperature for aluminum substrate of high power LED without window setting up in the Venturi is 89.8 °C. showed Fig. 5.

3 Experiment Results and Discussion a. The comparison of window type Venturi and without window on the effect of substrate heat dissipation for high power LED headlamp of 85 W Based on the experiment results, the highest temperature for the window type Venturi reached at 74.6 °C while the one without window setting came up to 96.4 °C. The effective heat dissipation temperature was 21.8 °C, cooling effect 22.61%. The result showed the window type Venturi had better heat sink performance. In Fig. 6 showed the comparison of temperature trend for both types of Venturi by ﬁnite element analysis.

120

85W LED has not venturi tube heat dissipaƟon 85W LED has venturi tube heat dissipaƟon 85W LED F.E.M. simulaƟon has venturi tube heat dissipaƟon

Temperature (oC)

100

80

60

40

20

0 1

5 10 20 30 40 50 60 61 62 63 64 65 66 67 68 69 70 71

Time (min)

Fig. 6. The temperature trends for window type and without window Venturi of high power LED headlamp of 85 W by ﬁnite element analysis

b. The comparison of window type Venturi and without window on the effect of substrate heat dissipation for high power LED headlamp of 80 W. Based on the experiment results, the highest temperature for the window type Venturi reached at 73.7 °C while the one without window setting came up to 93.3 °C. The effective heat dissipation temperature was 19.6 °C, cooling effect 21%. The result showed the window type Venturi had better heat sink performance. In Fig. 7 showed the comparison of temperature trend for both types of Venturi by ﬁnite element analysis.

An Innovative Design and Simulation of Transom Type Venturi Cooling Design

427

100

80W LED has not venturi tube heat dissipation 80W LED has venturi tube heat dissipation 80W LED F.E.M. simulation has venturi tube heat dissipation

90

Temperature (oC)

80 70 60 50 40 30 20 10 0 1

5 10 20 30 40 50 60 61 62 63 64 65 66 67 68 69 70 71

Time (min)

Fig. 7. The temperature trends for window type and without window Venturi of high power LED headlamp of 80 W by ﬁnite element analysis

c. The comparison of window type Venturi and without window on the effect of substrate heat dissipation for high power LED headlamp of 100 W in serial and parallel form. Based on the experiment results, the highest temperature for the window type Venturi reached at 64.8 °C while the one without window setting came up to 89.8 °C. The effective heat dissipation temperature was 25 °C, cooling effect 27.83%. The result showed the window type Venturi had better heat sink performance. In Fig. 8 showed the comparison of temperature trends for both types of Venturi by ﬁnite element analysis.

100 90

100W LED has not venturi tube heat dissipation 100W LED has venturi tube heat dissipation 100W LED F.E.M. simulation has venturi tube heat dissipation

Temperature (oC)

80 70 60 50 40 30 20 10 0 1

5 10 20 30 40 50 60 61 62 63 64 65 66 67 68 69 70 71

Time (min)

Fig. 8. The temperature trends for window type and without window Venturi of high power LED headlamp of 100 W in serial/parallel form by ﬁnite element analysis

428

M.-T. Sheen and Q. Wang

4 Conclusion 1. The high power LED substrate coated with aluminum nitride (AlN) has heat sink effect. The lamp holder without window setting in the Venturi reached 96.4 °C for LED of 85 W, 93.3 °C for LED of 80 W, and 89.8 °C for LED of 100 W in serial parallel form. 2. The records embedded with high power LED of 80 W, 85 W showed the current of 400 mA, voltage of 38.89 V, lumen of 6184 ml, 6484 ml, and color temperature of 5680 K. By static measurement for the highest temperature of LED lamp holder without window setting in the Venturi reached 93.3 °C, 96.4 °C. The same case for the lamp holder with window type Venturi came up to temperature of 73.7 °C, 74.6 °C, and 41.8 °C, 46.4 °C for the lateral edge. By compared with both situations, the effective heat dissipation temperature was 19.6 °C, 21.8 °C, cooling effect 21%, 22.61%. The results came from ﬁnite element simulation and coincided with the theory very much. 3. The records embedded with high power LED of 100 W in serial and parallel form showed the current of 400 mA, voltage of 26.01 V, lumen of 5462 ml, and color temperature of 5851 K. By static measurement for the highest temperature of LED lamp holder without window setting in the Venturi reached 89.8 °C. The same case for the lamp holder with window type Venturi came up to temperature of 64.8 °C, and 40.3 °C for the lateral edge. By compared with both situations, the effective heat dissipation temperature was 25 °C, cooling effect 27.83%. The results came from ﬁnite element simulation and coincided with the theory very much. From the results of experiment and theory showed that the window type Venturi lamp holder is an effective device for heat dissipation and could be used to handle the light failure problem for LED. The Venturi with window type has a sound mechanism of convection for heat sink and is potential for LED products in the market. The substrate of LED is connected with the lamp holder to conduct the heat and cooled down by the natural convection from the air through the window in the Venturi. The experiment evidences showed that the window type Venturi is truly good at heat dissipation and can achieve the goal of energy and carbon saving. Acknowledgments. This work was supported by The Ministry of Industry and Information Technology (2016-755-63)

References 1. Gu, Y., Narendran, N., Freyssinier, J.P.: White LED performance. In: Proceedings of SPIE, vol. 5530, pp. 507–514 (2004) 2. Park, J., Shin, M., Lee, C.C.: Measurement of temperature proﬁles on visible light-emitting diodes by use of a nematic liquid crystal and an infrared laser, Optics Lett. 29, 2656–2658 (2004) 3. Niki, I., Narukawa, Y., Morita, D., Sonobe, S.Y., Mitani, T., Tamaki, H., Murazaki, Y., Yamada, M., Mukai, T.: White LEDs for solid state lighting. In: Proceedings of SPIE, vol. 5187, pp. 1–9 (2004)

An Innovative Design and Simulation of Transom Type Venturi Cooling Design

429

4. Shatalove, M., Chitnis, A., Yadav, P., Hasan, M.F., Khan, J., Adivarhan, V., Maruska, H.P., Sun, W.H., Khan, M.A.: Thermal analysis of flip-chip packaged 280 nm nitride-based deep ultraviolet light-emitting diodes. Appl. Phys. Lett. 86, 201109 (2005) 5. Cheng, Q.: Thermal management of high-power white LED package. In: IEEE, 8th International Conference on Electronic Packaging Technology, Shanghai, China, pp. 1–5, 14–17 August 2007 6. Zhang, Z.: LED car headlights of the heat analysis. The National University of Successful Master’s Thesis (2008) 7. Robert, S., Metz, E., Setauket, N.Y.: Heat-dissipating method and device for LED display. United States Patent 5,173,839, 22 December 1992 8. Hochstein, P.A., Troy, M.: LED Lamp Assembly with Means to Conduct Heat Away from the LEDs, United States Patent 6,045,240, 4 April 2000 9. Azar, K., Mandrone, C.D.: Effect of pin ﬁn density of the thermal performance of Unshrouded Pin Fin HeatSinks. In: Ninth Annual IEEE (1995) 10. Li, J., Yanga, Q., Niu, P., Jin, L., Meng, B., Li, Y., Xiao, Z., Zhang, X.: Analysis of thermal ﬁeld on integrated LED light source based on COMSOL multi-physics ﬁnite element simulation. Physics Procedia 22, 150–156 (2011)

The Study of Predictive Model for ZD6 Switch Current Based on Wavelet Neural Network Junfeng Zheng1(&), Hanyue Zhao1, Jie Zhang2, and Yunlong Li1 1

2

Hefei Metro Co. Ltd., Hefei 230000, China [email protected] School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China

Abstract. High-speed switch is a key infrastructure that affects the safety of high-speed railways. Under the action of train and temperature load, the switch will continue to undergo damage deformation, which is the key and difﬁcult point in line maintenance. Once a switch fails, it is very likely to cause a major safety incident. This paper mainly uses the general ZD6 switch current as the research object. First, analyze the normal curve of the ZD6 switch current, sum up the common faults, analyze the causes of the faults and the phenomena that occur. For the analysis and prediction of the switch current, a method of establishing a wavelet neural network model is adopted. Keywords: ZD6 switch

Network model of wavelet neural Predict current

1 Switchcurrent Curve 1.1

The Conditions of Signal Centralized Monitoring System

In 2010, the directive No. 709’2010 Technical Conditions for Signal Microcomputer monitoring System promulgated by the ministry of Railways Transportation Authority stipulated the system structure, monitoring content and technical requirements of the railway signal centralized monitoring system. Among them for the DC switch machine, the monitoring content has the action current, fault current, action time and conversion direction of the switch during the conversion process: the monitoring point is the action return line: the monitoring range: current 0–10A (single machine), The time is not more than 0.1 s; test mode: continuous test according to IDQJ condition; sampling rate: 40 ms. 1.2

The Switch Current Curve of ZD6

The ZD6 switch action current curves monitored at 10 time points are obtained respective1y with the MATLAB to draw the graph. Only a set of current curves are listed in this paper, as shown in Fig. 1, which is 22:25:56 on July 30, 2011. The action current curve of the ZD6 switch from the reverse position to the positioning. Analysis Fig. 1, the switch current curve conforms to the normal switch current curve, 1 is the unlocking zone, the current is large when the motor starts up initially, © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 430–439, 2019. https://doi.org/10.1007/978-3-030-04582-1_50

The Study of Predictive Model for ZD6 Switch Current

431

2

1.5

Currentvalue

1

0.5

0

0

1

2

3

4

5

6

Time/s Fig. 1. ZD6 switch action current curve

and the unlocking process is completed by the spindle rotation; 2 is the action zone, after the switch is unlocked, the rack block moves and drives The switch machine moves to the speciﬁed position; 3 is the lock area, after the tiprail is in place, the start circuit is disconnected the switch is locked, and the action bar is prevented from reversing under the action of external force; 4 is the slow release zone, after the switch islocked, the curve A straight line of approximately zero will appear on it. The remaining 9 groups of switch currents were plotted, and the9 sets of switch current curves were normal.

2 Wavelet Network Model 2.1

Model Establishment

The wavelet network model is designed according to the characteristics of the ZD6 switch current, which is divided into an input layer, an implicit layer and an output layer. The input layer input is the switch current of the ﬁrst n time points of the current time point; the hidden layer node is composed of a wavelet function; and the output layer output is the predicted switch current of the current time point. 2.2

Data Scheme

The neural network prediction process is s actually a process of continuously learning and memorizing the training samples, and then performing l the same regular output on the test samples. Therefore, the selection of training samples and test samples is particularly important. When building a network, you need to divide the input data into three parts: training samples, test samples, and test samples. Due to the complexity of the ZD6 switch and the learning characteristics of the wavelet network, the trend prediction of

432

J. Zheng et al.

the wavelet network can be realized only when the sample information contained in the training sample is sufﬁciently comprehensive [1]. In this paper, two data schemes are selected. The ﬁrst scheme is that the input sample data is 4 sets of different currents and the same monitored current at the same time; and the output sample data is the ﬁfth set of current sample data. There are 140 pairs (input and output) sample pairs. The current sample data is shown in Table 1. The second scheme is that the current sample data is based on a few days before the current time to predict a part of the data after the current time. There are 588 (input and output) sample pairs, some current samples. The data is shown in Table 2. Table 1. Part of the current sample data of the scheme Input sample 1.196 1.922 1.608 1.843 1.137 1.294 0.824 0.941 0.667 0.745 0.588 0.627

Expected output 1.765 1.843 1.294 0.941 0.745 0.627

1.961 1.451 1.02 0.784 0.667 0.588

1.804 1.882 1.333 0.941 0.745 0.627

The input sample data of scheme 1 is shown in Fig. 2, the output sample data is shown in Fig. 3: the input sample data of scheme 2 is shown in Fig. 4, and the output sample data is shown in Fig. 5. 2.3

Network Structure

In this paper, the number of hidden layers in the wavelet network is set to 1. According to the structure of the wavelet network, the following is a detailed introduction to analyze the network structure according to the number of input layer nodes, the number of hidden layer nodes, the number of output layer nodes, the original data processing, and the selection of the conversion function Number of input layer nodes: As can be seen from the previous sample data structure, the operating currents of the ﬁrst four time points of the current time point are input as a network, so the number n of neurons- in the input layer is set to 4. Number of hidden layer nodes: The number of hidden layer neuronsisnot calculated by a clear formula. The number of hidden layer nodes can only be determined by a combination of trial and error methods and empirical formulas. Number of output layer nodes: The predicted voltage value at the current time point is output, so the number m of neurons in the output layer is set to 1. Raw data processing: Due to the special requirements of the neural network for input data, in order to speed up the convergence of network training the input data needs to be normalized, that is that the input and output data need to be mapped to [−1,1]. In MATLAB, the mapminmax function is used to normalize and denormalized the input data of the network, and the output data of the network is denormalized.

The Study of Predictive Model for ZD6 Switch Current Table 2. Scheme 2 Partial Current Sample Data Input sample 1.961 1.608 1.608 1.137 1.137 0.824 0.824 0.627 0.667 0.588 0.588 0.549

Expected output 1.137 0.824 0.667 0.588 0.549 0.51

0.824 0.667 0.588 0.549 0.51 0.51

0.667 0.588 0.549 0.51 0.51 0.471

2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4

0

20

40

60

80

100

120

140

Fig. 2. Scheme 1 input sample data

2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4

0

20

40

60

80

100

120

Fig. 3. Scheme 1 output sample data

140

433

434

J. Zheng et al. 2

1.5

1

0.5

0

0

100

200

300

400

500

600

Fig. 4. Option2input sample data

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

0

100

200

300

400

500

600

Fig. 5. Scheme 2 output sample data

Equation 2-1 is the calculation formula for the mapminmax function. xi ¼

xi xm in xmax xm in

ð2 1Þ

Where the data used to represent the output or input is xi, the minimum used to represent the change in data is xmin, the maximum value used to represent the change in data is xmax. Conversion function selection: The implicit layer conversion function of the wavelet network is composed of wavelet basis functions. The wavelet function can extract the sample set and discover the dataset law and store it. The wavelet basis function selects the most commonly used Morlet mother wavelet basis function, and its formula is shown in 2-2. y ¼ cosð1:75xÞex

2

=2

The mother wavelet function waveform is shown in Fig. 6.

ð2 2Þ

The Study of Predictive Model for ZD6 Switch Current

435

1 0.8 0.6

y

0.4 0.2 0 -0.2 -0.4 -6

-4

-2

0 x

2

4

6

Fig. 6. Waveform of the mother wavelet function

In the wavelet neural network, x in the above ﬁgure is the value of the input layer input multiplied by the weight; y is the value calculated by the wavelet basis function, which will be the output of the hidden layer node [2]. Error adjustment: the error back propagation process in the BP network learning process refers to the process of minimizing the sum of the squares of the error between the actual output and the desired output. The wavelet neural network weight parameter weight correction algorithm is similar to the BP neural network weight correction The algorithm is usually done using the gradient descent method. The standard gradient correction method is to perform an error correction on the weight for each input and output training of the sample. This method is easy to fall into local error when training the network. Therefore, in order to solve this problem, this paper uses global error adjustment. The so-called global error adjustment refers to calculating the cumulative error sum after all the learning samples are systematically studied, and then modifying the weight. The global error adjustment method [3] used in the error adjustment process of the wavelet network in this paper. Other parameters: The accuracy of neural network prediction is also high. It is also related to the network learning rate and the number of network iterations. The learning rate ranges from 0.01 to 0.8. In this paper, 0.01 is selected. The larger the network iteration, the smaller the prediction error is, but it is too large. The efﬁciency of the algorithm will be greatly reduced. This article chooses 100.

436

J. Zheng et al.

2.4

Algorithm Flow

In this paper, wavelet neural network is chosen as the method of ZD6 ballistic motion prediction. The network training steps of the wavelet neural network algorithm are introduced [4] in detail as follows: Step1: Network parameter initialization The number of input layer nodes n and the number of output layer neurons m are determined according to the characteristics of the current data samples; the number h of hidden layer neurons is determined according to an empirical formula; the network learning rate g and the number of network iterations are determined by analyzing network errors. Using random number generation function randn() the wavelet function scaling factor aj the translation factor bj and the network weight xij , xij , are initialized. Step2: Data sample selection Import the monitored historical data and divide the data into training samples and test samples: use training samples for network training; use test samples to test network prediction accuracy. Step3: Data preprocessing Data samples are normalized to meet neural network input requirements. Step4: Error calculation The training samples are imported into the wavelet neural network, the input is multiplied by the conversion function and the weight, the network prediction output is obtained, and the value is compared with the expected output of the network to calculate the error between the network prediction output and the expected output e. Step5: Weight correction Using the gradient descent method, the network weight and wavelet function parameters are continually corrected according to the error and ﬁnally the network prediction value is continuously approached to the expected output value [5]. Step6: Determine whether the algorithm can end. If not, return to step 4. Step7: Import the test sample and normalize it. The test data is input to the trained network to determine the predicted result. Step 8: Perform error analysis on the prediction results and end.

3 Simulation Results and Error Indicators After the network structure is determined, the network training is performed, and then the time series subjected to the prediction processing is substituted into the prediction equation to obtain a predicted value. The data dynamic: scheme-trained wavelet neural network is used to predict the ZD6 switch action current. The comparison between the prediction result and the actual switch action current is shown in Fig. 7. The data channel scheme 2 trained wavelet neural network is used to predict the ZD6 switch action current and predict. The result is compared with the actual switch action current as shown in Fig. 8.

The Study of Predictive Model for ZD6 Switch Current

437

Prediction of switch current 2 Predictive current value Actual current value

1.5

Currentvalue

1

0.5

0

0

20

40

60

80

100

120

140

Time

Fig. 7. Prediction result one

Prediction of switch current 2 Predictive current

1.8

Actual current

1.6 1.4

Currentvalue

1.2 1 0.8 0.6 0.4 0.2 0

0

20

40

80

60

Time

Fig. 8. Prediction result two

100

120

140

438

J. Zheng et al.

To judge the accuracy of the predicted value, it is necessary to make an error analysis on the predicted value to judge the quality of the model. For the neural network, the following two indicators can be used to analyze the error: nP þ m x^ x i i 1 1. Average relative percentage error(MAPE) MAPE ¼ m ð xi 100%Þ, the i¼n þ 1

smaller the value, the better. 2. Directional statistical indicator (Dstat), the larger the coefﬁcient, the better the prediction effect of the model. The calculation formula is shown in 3-1 and 3-2. 1 Xl a n¼1 n N

ð3 1Þ

1 ðxn þ 1 xn Þð^xn þ 1 xn Þ [ 0 0 other

ð3 2Þ

Dstat ¼ an ¼

Where N is the size of the test data set, 1 is n − 1, xn þ 1 is the actual value time series value, and ^xn þ 1 is the predicted value. Two indicators can be obtained by programming in MATLAB. The forecasting indicators of the two schemes ares shown in Table 3. Table 3. Forecast indicators

Option One Option Two

Average relative percentage error (MAPE) 8.77%

Directional statistical indicator (Dstat) 0.73

5.26%

0.31

It can be seen from Table 3 that the wavelet neural network can predict the ZD6 switch action current [6] more accurately, and the network prediction value is close to the expected value.

4 Conclusion Artiﬁcial neural network technology has strong robustness and fault tolerance, as well as certain promotion generalization ability and self-learning ability. It is especially suitable for solving the problem of non-deterministic judgment, reasoning and prediction of complex causality, which can solvethe switch current. Forecast the problem. Acknowledgement. This research was supported by the National Key R&D Program of China (2017YFB1200702), National Natural Science Foundation of China (Project No. 61703351), Science and Technology Plan of Sichuan province (Project No. 2017ZR0149, 2017RZ0007, 2017015, 2018RZ0078), Science and Technology Plan of China Railway Corporation (Project

The Study of Predictive Model for ZD6 Switch Current

439

No.: 2016X006-D), Chengdu Soft Science Research Project (2017-RK00-00028-ZF, 2017RK00-00378-ZF) and the Fundamental Research Funds for the Central Universities (2682017CX022, 2682017CX018).

References 1. Li, Z.: Analysis of the action current curve of switch. Railway Commun. Signal 41(11), 20–21 (2005) 2. Gu, J.: Analysis and treatment of several common switch fault action current curves. Railway Oper. Technol. (1), 11 2014 3. Dai, S.: Failure analysis and preventive measures of ZD6 electric switch machine. Shanghai Railway Sci. Technol. (2004) 4. Chao, L.: Analysis and improvement measures of friction current imbalance for ZD6 electric switch machine. Railway Signal. Commun. (1995) 5. Zhao, Y.: Adjustment method of ZD6 electric switch machine. Railway Signalling & Communication, July 2002 6. Lin, Y.: Railway Station Signal Interlocking. China Railway Press, Beijing (2007)

PCB Image Registration Based on a Priori Threshold SURF Algorithm Jing Huang1,2(&), Junnan Li1,2, Lisang Liu1,3, Kan Luo1,2, Xiaoyong Chen1,2, and Fenqiang Liang1,2 1

Fujiang University of Technology, Fuzhou 350118, Fujian, China [email protected], [email protected], [email protected], [email protected], [email protected], [email protected] 2 Fujian Research Base of Industrial Integration Automation Industry, Fuzhou 350118, Fujian, China 3 Fujian Key Laboratong of Automotive Electronics and Electric Drive Technology, Fuzhou 350118, Fujian, China

Abstract. In order to solve the problems of the traditional speed up robust features (SURF) algorithm, which is used in PCB board defect detection, such as mis-match, low precision, and slow speed, this paper proposes a SURF algorithm based on error prior information for PCB image matching. First, the feature points are extracted by the SURF algorithm and the distances of the feature point pairs are calculated. Then, according to the prior information of the mechanical error of the PCB motion platform, the false matching pair is eliminated based on the a priori threshold boundary condition. Finally, the PCB images are registrated based on the mapping relationship between the images satisfying the least squares ﬁtting criterion. Experiments show that the algorithm proposed in this paper has higher accuracy and faster matching speed, which is conducive to the improvement of PCB board defect detection efﬁciency. Keywords: Image registration SURF PCB

Feature extraction A priori threshold

1 Introduction In recent years, PCB vision detection based on machine vision has become a hot issue in the industrial ﬁeld. PCB image registration is a key step in machine vision detection of PCB defects. The registration algorithm can directly affect the speed and accuracy of the entire inspection system [1]. At present, there are three main types of image registration algorithms: (1) Grayscale based matching, (2) Feature-based matching [2, 3], (3) Image matching based on understanding [4, 5]. Among the researches that using the ﬁrst type algorithm, the images are matched directly by the gray scale of the image without extracting image features, such as minimum distance method, cross correlation method, correlation coefﬁcient method, etc. [6]. Although such methods have high registration accuracy, their calculation speeds are slow. Among the researches that using the second type © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 440–447, 2019. https://doi.org/10.1007/978-3-030-04582-1_51

PCB Image Registration

441

algorithm, it is necessary to extract image features (such as feature points, lines, faces, textures, etc.) and then match the images. This method greatly compresses the amount of image information, and the calculation speed is fast [7]. The accuracy of registration depends on the accuracy of feature point extraction and matching [8]. For example, a fast matching search algorithm based on image gray value feature points is proposed in the literature [7], which changes the search strategy of the traditional matching method argotic properties. The researches using the third type algorithm are still in a preliminary discussion stage. In summary, to match images by their feature points is a commonly used and easier method, which is the mainstream of current research for its guarantee of the resolution, rotation, translation, etc. In view of the existing problems of the algorithms, combined with the actual requirements for the registration speed and accuracy of the PCB image, an improved SURF algorithm based on distance screening is proposed in this paper. According to the prior information of the mechanical error of the PCB motion platform, the mismatching pair is eliminated by setting the priori threshold boundary condition. Then the image registration experiments are carried out on a PCB with mechanical motion error of 0.05–0.1 mm and size of 42 * 42 mm. The experimental results show that the image registration method proposed in this paper is faster and more accurate, and is suitable for PCB defect detection in production lines.

2 Traditional SURF Algorithm 2.1

SURF Feature Point Extraction

In the traditional SURF algorithm, the Hessian matrix is used for feature point extraction and the integral image is used for convolution operations to achieve fast registration. There are three steps to extract the feature points: (1) SURF feature point detection: Assuming that f ðx; yÞ is the gray values of each pixel in the image, the matrix H is obtained by ﬁnding the second-order partial derivative for each direction. Its Hessian matrix is: " Hðf ðx; yÞÞ ¼

@2 f @x2 @2 f @x@y

@2f @x@y @2f @y2

# ð1Þ

@ f @ f Where x, y represent the pixel coordinates, respectively; @x 2 and @y2 are the secondorder partial derivatives of the pixels in the x, y direction of the corresponding positions @2f in the image, respectively. @x@y is the second-order partial derivative of the pixel in the x, y direction at the corresponding position in the image. 2

2

442

J. Huang et al.

In order to make the feature have rotation and scale invariance, the algorithm performs scale space transformation (convolution operation) on the image by constructing an approximation instead of Gaussian second derivative BOX-Filter (box ﬁlter), whereby the Hessian matrix becomes: H¼

Dxx Dxy

Dxy Dyy

ð2Þ

Where Dxx , Dxy , and Dyy represent convolutions in three directions, and Dxy and Dyy have the same meaning. Thus the Hessian matrix is simpliﬁed as: detðHÞ ¼ Dxx Dyy ð0:9Dxy Þ2

ð3Þ

In the above formula, 0.9 is the normalized ratio, and the value of D in the formula can be quickly calculated by integrating the image [9], thereby obtaining the eigenvalue of the pixel in the image. As shown in Fig. 1, the points are detected on different scales (taking the pixel points marked as crosses as an example), and the “spot response map” at different scales is established by changing the size of the ﬁlter. By judging the feature value of the pixel, it is possible to determine whether it is a feature point or not. For example, the determinant of the Hessian matrix is positive, then the feature points therein are local maximums of the 3 3 3 neighborhood.

Fig. 1. The diagram of least error principle

SURF feature point direction determination. In order to make the feature points have rotational invariance, it is necessary to determine the main direction of the feature points. The Haar wavelet response of the feature points in the x, y direction is solved in the circular region with the feature points determined above as the center, and the dx, dy coordinate system is established. By counting the sum of all Haar wavelet responses, the longest vector direction is the main direction. Find the feature point description operator. After the direction is determined, a region of 20 s (s is the scale corresponding to the feature point) is constructed centering on the feature point, and the region is divided into smaller (4 4) square sub-regions.

PCB Image Registration

443

Then, for each sub-area, (5 5) spatially normalized sampling points are calculated, and the Haar wavelet response dx, dy of each sampling point is calculated. The Gaussian kernel is then used to weight the dx, dy of P each subdomain P P so that P each subﬁeld obtains the four-dimensional feature vectors dx, dy, jdxj, jdyj. Finally, a (4 4 4) 64-dimensional SURF description operator is constructed. By normalizing the descriptors, the descriptors have scale invariance. 2.2

SURF Feature Point Matching

After the feature points of the reference image and the image to be detected obtained by the SURF algorithm, feature matching is performed. The traditional SURF algorithm feature point matching is to obtain the distance between the feature point of the image to be detected and the adjacent feature point of the reference image ﬁrstly, and then obtain the distance between the feature point of the image to be detected and the next adjacent feature point of the reference image. Finally, the distance ratio is used to determine the degree of matching between the feature points. If the ratio is less than a certain threshold T, the match is considered successful. Assuming that the feature point pairs are x, y, the Euclidean distance is deﬁned as: vﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ u 64 uX dðx; yÞ ¼ t ðxi yi Þ2

ð4Þ

i¼1

Firstly, calculating the Euclidean distance d1 of the feature point xi of the image to 0 be detected and the nearest feature point yi in the reference image; Then, calculating the Euclidean distance d2 of the feature point xi to be detected and the next closest feature point y00i in the reference image; Finally, the ratio calculation is performed, the ratio r = d1/d2, and if it is less than 0 the preset threshold T, then xi matches yi , and vice versa.

3 Improved SURF Feature Point Matching The traditional SURF algorithm performs 2 distance calculations, 1 time ratio calculation, and ﬁnally compares the ratio with the threshold. The calculation is cumbersome and may lead to mismatch. Combined with the actual situation of PCB defect detection, under the a priori condition of known mechanical system error range, this paper proposes an improved SURF algorithm combined with distance screening. That is, the SURF feature points are determined by the Euclidean distance and the set mechanical error range as the screening conditions.

444

J. Huang et al.

Assuming Aðx1 ; y1 Þ and Bðx2 ; y2 Þ are a pair of the feature points, the Euclidean distance is deﬁned as: dðA; BÞ ¼

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ðx1 x2 Þ2 þ ðy1 y2 Þ2

ð5Þ

The matching is performed by calculating the Euclidean distance between each potential feature point pair. In this step, there are many feature point pairs, which usually contain many suspicious matches. By setting the mechanical error range ½Dmin ; Dmax , the mismatch is eliminated. If the Euclidean distance of the feature point pair satisﬁes the range, the match is successful, otherwise the match fails. As shown below: Dmin \dðx; yÞ\Dmax

ð6Þ

The speciﬁc steps of feature point matching are as follows: (1) Extracting feature points from the reference PCB image by the SURF algorithm; (2) Converting the known mechanical range of the motion platform into a pixel error range by calibration; (3) Acquiring an image to be detected and acquiring a SURF feature point of the image to be detected; (4) Matching the feature points of (1) and (3) by the SURF algorithm; (5) Calculate the Euclidean distance of the matched feature point pairs in (4); (6) Comparing the Euclidean distances of the matched feature point pairs in (5) by the pixel error range obtained in (2). If the Euclidean distance satisﬁes the threshold range of the pixel error, it is a valid matching point, and vice versa;

4 Experimental Results and Analysis The experiments were carried out on a desktop industrial computer (i7 4700, 8G memory) with the MTLAB software system installed. The acquisition system hardware includes MQ042CG-CM industrial camera, HK3514MP5 35 mm C-port lens, selfmade flat color tunable light source, and AMC4030XY axis servo stepping motion control platform. The pixel of the experimental PCB image is 580 * 580. With a mechanical motion error range of 0.05–0.1 mm (Dmin ¼ 0:05mm, Dmax ¼ 0:1mm) as a priori condition, a mechanical error threshold parameter T corresponding thereto is set. In this paper, the upper limit of T is set to 450, (since the lower threshold is very small, this is equivalent to 0). The size of the PCB to be tested in this experiment is 42 * 42 mm, and 100 feature point matching experiments are performed. The experimental results were evaluated by four measures of Euclidean distance standard deviation, variance, calculation speed and mutual information.

PCB Image Registration

4.1

445

Image Matching Effect Experiment

In the case that the camera parameters are ﬁxed and the motion control platform error range is known, it can be seen that the conventional SUFR algorithm has multiple pairs of mismatched features, and accurate image alignment cannot be achieved. The Euclidean distance range of the matching feature point pair distribution is obtained by statistics. It can also be seen that compared with the traditional feature point matching algorithm, the feature point pair distance after the matching by the algorithm is within the normal range and limited deviation. The standard deviation and variance of the corresponding distances obtained are shown in Table 1. It can be seen that the improved SURF matching algorithm used in this paper is much smaller than the traditional algorithm in terms of variance and standard deviation, and the average mutual information becomes larger and the precision is improved. Table 1. Comparison of 100 sets of PCB registration results

SURF Improved SURF

4.2

Standard deviation of feature point distance 3.5502e + 04 1.1204e + 02

Variance of feature point distance 1.2604e + 09 1.2555e + 04

Average mutual information 0.1199 2.9540

Feature Point Matching Positioning Experiment

Firstly, the experimental PCB image (580 * 580 pixels) is horizontally shifted to the right by 2 pixels and vertically downward by 3 pixels. Then the positioning experiment based on SURF feature point matching algorithm is performed. The experimental results are shown in the following Table 2: Table 2. Positioning based on SURF feature point algorithm Improved type SURF

Improve SURF

Perspective transformation matrix H 2

3 0:5104 0:4309 240:5379 4 0:5728 0:4988 278:4644 5 0:0019 0:0016 1 2 3 1:0025 0:0006 2:3664 4 0:0009 1:0024 3:4238 5 0 0 1

Mutual information value 0.0708

Speed (ms) 496.748

2.7964

324.956

It can be seen from the table that the perspective transformation matrix generated by the traditional SURF algorithm is faulty and cannot be used for image localization. The error of the positioning accuracy of the algorithm is within 1 pixel, which satisﬁes the detection requirements of the system. Compared with the traditional SURF algorithm, the mutual information is obviously improved, the speed is increased by 171.828 ms, and the positioning efﬁciency is improved by 34.6%.

446

4.3

J. Huang et al.

Registration Defect Detection Experiment

In this paper, the detection map (mechanical error is 0.1 mm) is used to perform the feature point registration defect detection test using SURF algorithm and improved SURF algorithm respectively, which are shown in Fig. 2 above. Because the improved SURF algorithm adopts the error prior information, set the threshold constraint and delete the feature point pairs that do not satisfy the constraint, the defect of the PCB without the pad can be effectively identiﬁed after the registration and subtraction operations. However, the traditional algorithms cannot detect PCB defects. The experimental results show that the proposed constraints can effectively improve the matching accuracy of the algorithm (Fig. 3).

(a) SURF

(b) Improved SURF

Fig. 2. PCB feature point matching diagram

(a) SURF (failure)

(b) Improve SURF

Fig. 3. PCB registration defect detection results

5 Conclusion In view of the existing problems of the current SURF algorithm, combined with the actual PCB defect detection production line requirements, this paper proposes an improved SURF algorithm combined with distance screening. According to the prior information of the mechanical error of the PCB motion platform, the mismatch pair is eliminated by setting the priori threshold boundary condition. Experiments show that the algorithm effectively improves the matching accuracy and positioning speed, and is suitable for PCB defect detection in production lines. However, the algorithm still has shortcomings, such as the selection of a priori threshold, the relationship between the number of PCB components and the matching speed and accuracy, which need to be further studied.

PCB Image Registration

447

Acknowledgement. The authors would like to thank the anonymous reviewers for their valuable comments. This work was supported by Fujian Provincial Natural Science Foundation Projects (2012D069), Natural Science Foundation of Fujian Province (2018J01640), Fujian Provincial Education Department Youth Fund (JAT170367), Initial Scientiﬁc Research Fund of FJUT (GY-Z12079), Pre-research Fund of FJUT (GY-Z13018) and China Scholarship Council (201709360002).

References 1. Huang, J., Chen, X., Luo, K., Ji, Y., He, S., Lin, L., Li, J.: A modiﬁed SURF algorithm for PCB image registration. J. Fujian Univ. Technol. 16(03), 275–280 (2018) 2. Goshtasby, A.A.: 2D and 3D Image Registration: For Medical, Remote Sensing, and Industrial Applications. John Wiley & Sons Inc., Hoboken (2005) 3. Duan, C., Meng, X., Tu, C., et al.: How to make local image features more efﬁcient and distinctive. IET Comput. Vis. 2(3), 178–189 (2008) 4. Kolmogorov, V.: Graph Based Algorithms for Scenery-Construction from Two or More Views. The Graduate School of Cornell University (2004) 5. Lou, B.: Study on Key Techniques of Region based Image Matching Algorithm. XI DIAN University, Xian (2006) 6. Fan, L., Wang, Y., Gao, X.: A new gray - based method of images matching. Microcomput. Inf. 30, 296–297+8 (2007) 7. Gan, J., Wang, X., Quan, W.: A fast image matching algorithm based on characteristic points. Electr. Opt. Control 16(02), 64–66 (2009) 8. Barrera, F., Lumbreras, F., Sappa, A.D.: Multispectral piecewise planar stereo using Manhattan-world assumption. Patt. Recogn. Lett. 34, 52–61 (2013) 9. Yin, J., Cheng, L., Fan, M., et al.: Study of PCB components recognition based on SURF. School Electr. Inf. Eng. Hefei Normal Univ. 33(06), 13–16 (2015)

The Calculation of Safety Front Boundary of Paired Approach Procedure Based-on Escape Maneuver X. He1, F. Zhang1(&), J. Chen1, and F. Song2 1

School of Air Trafﬁc Management, Civil Aviation Flight University of China, Guanghan 618307, China [email protected], [email protected], [email protected] 2 Operation and Control Department, Shanxi Branch of China Eastern Airlines, Taiyuan 030031, China [email protected]

Abstract. The paper studying the case that blunder intrude the approach path of lead aircraft in paired approach, and advances three escape maneuver procedure by introducing the paired approach. This paper gives ﬁve steps to determine the safety front boundary by establishing relative motion reference system. At last, the paper takes a case to make the simulation and calculate the minimum safety front boundary between different lateral spacing. The results show that distance between runway centerlines and the intrusive angle of intrusive plane have effect on minimum follow-up distance. Meanwhile, this method can calculate the safety front boundary of paired approach in real time and quantitatively. Keywords: Paired approach Escape maneuver procedure Safety front boundary Lateral spacing

1 Introduction With the continuous expansion of the demand for civil aviation transportation in China, more and more airports in China use multi-runway operation. It is designed to cope with the fast-growing air trafﬁc. The closely spaced parallel runways are widely used because of its irreplaceable advantages. The paired approach mode, as a new technology applied to the closely spaced parallel runways, has a signiﬁcant effect on the improvement of runway capacity. At present, China’s airport and airborne equipment conditions can meet its operational requirements, but the running interval of the paired approach lacks the corresponding aviation regulations, so it is necessary to conduct an in-depth study of its safety area.

Research supported by 2015 Civil Aviation Science and Technology Innovation Guide Project (20150226). Innovation Team Support Project Plan (XM2628). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 448–456, 2019. https://doi.org/10.1007/978-3-030-04582-1_52

The Calculation of Safety Front Boundary of Paired Approach Procedure

449

The NASA Langley Research Center ﬁrst proposed the concept of paired approach in 1996, and J Hammer elaborated on it and analyzed the influencing. Factors later [1, 2]. In 2000, Bryant T. and James K. established the collision risk and wake diffusion model for two aircraft in the air, pointing out that in the approach phase, avoidance aircraft climbing and turning quickly is the most effective way to resolve flight conflicts [3]. Teo and Tomlin studied the methods and procedures for paired approaches to the danger zone and performed a scenario simulation to derive the minimum runway interval for the paired approach within an acceptable level of safety [4–6]. In 2011, R R. Eftekari and other scholars calculated the safety interval values under Class I and Class II instrument landing standards and different runway conﬁgurations [7]. In 2013, NASA proposed a new concept of paired approach, and the rear aircraft can surpass the front in the safe range [8, 9]. In China, Lu Fei et al. constructed a longitudinal collision risk model based on wake avoidance and positioning error, and concluded that the shorter the distance between the paired front and the runway entrance, the greater the collision risk [10]. In 2013, Niu Xialei et al. constructed a minimum following distance model by studying the movement of paired aircraft over time during the paired approach [11]. In 2015, Tian Yong and others took into account the influence of the error of the lead aircraft and the influence of the wake turbulence on the rear aircraft, and the optimal initial longitudinal interval between the two aircraft was obtained [12]. In 2016, Han Dan et al. divided the paired approach procedure into the rear aircraft to not surpass the lead aircraft and the rear aircraft to surpass the lead aircraft for comparison calculation, and obtained the minimum longitudinal interval in these two cases, thus determining the whole safety area of the paired approach procedure [13, 15, 16]. This paper considers the situation of miss approach, and analyzes the minimum interval requirements of the two aircraft in the ﬁnal approach phase of the paired approach. Then this article studies the calculation method of the anti-collision safety front boundary under the motorized avoidance procedure. Finally, the escape maneuver state determines the minimum longitudinal safety interval between the paired rear and the front.

2 Paired Approach Procedure The paired approach is a procedure similar to the related parallel instrument approach implemented in the closely spaced parallel runways under instrument meteorological conditions. The core idea is that the two aircraft should work closely together in the approach procedure, keeping one after the other. Rear aircraft must maintain two safe distances from the lead aircraft: one is to ensure the minimum following distance from the lead aircraft. When the lead aircraft mistakenly enters the rear track, the rear aircraft has enough time to deal with this situation. The other is to ensure that the distance between the lead and rear aircraft cannot exceed the maximum safety interval in the safety zone of the wake, that is, prevent the wake generated by the lead aircraft from affecting the safe flight of the rear aircraft, as shown in Fig. 1. In addition, in the existing airport navigation facilities, the implementation of the paired approach also requires the use of automatic dependent surveillance-broadcast (ADS-B) and cockpit

450

X. He et al.

display of trafﬁc information (CDTI) to establish and maintain the longitudinal safety interval between the two aircraft.

Fig. 1. Illustration of paired approach

3 Collision Avoidance Maneuvers and Safety Front Boundary In fact, the positioning of the receiving device fails, the pilot overcorrection after the gust of wind or the pilot’s right or wrong runway and other factors will cause the aircraft to break into error. In this case, we can think that the heading of the aircraft will escape at a greater angle not than a 60-degree turn. Therefore, it is sufﬁcient to select a turning angle of less than 45° in the emergency maneuver avoidance procedure. During the paired approach, different Emergency Escape Maneuver (EEM) needs to be preset according to the possible situation. Only the avoidance of the aircraft on the right side is considered [17]. Since a single emergency maneuver avoidance procedure does not satisfy most situations, this paper advances three maneuver avoidance procedures: the ﬁrst is that avoidance aircraft accelerate the straight line escape procedure; the second is that avoidance aircraft accelerate while turning to 45° straight line to escape; the third is that avoidance aircraft maintain the original speed to escape 60°. As shown in Fig. 2, this paper determines the safety front of the intrusion area by the following ﬁve steps: (1) Consider the path of the avoidance aircraft as a straight line segment plus an arc segment and determine an emergency maneuver avoidance procedure. (2) The relative positional relationship when the two aircraft are near-miss is used as the termination condition of the calculation. (3) Input the initial motion parameters when the aircraft intrude. (4) According to the initial parameters and the two-aircraft relative motion model, the actual flight trajectory of the intrusion aircraft is calculated. (5) Determine the anti-collision safety front boundary of the paired aircraft.

The Calculation of Safety Front Boundary of Paired Approach Procedure

451

Fig. 2. Five steps to determine the safety front

In reference [11], the coordinate system model of the initial motion state of the aircraft is established. This paper studies the situation of the wrong approach, establishes the motion model of avoidance aircraft and intrusion aircraft in the case of maneuver avoidance. As shown in Fig. 3, the relative position of the aircraft is represented by xi and yi , the lower corners i mean different aircraft; b indicates intrusion aircraft. Let e represent avoidance aircraft. Let bi represent the heading of aircraft i, which is obtained by rotating counterclockwise from the shaft. During the paired approach, both aircraft approach and land with a similar gradient glideslope. According to NASA’s deﬁnition of paired approach and the current operational practice in the United States, the two aircraft in the ﬁnal approach have no sufﬁcient safety intervals in the vertical direction before and after pairing. Meanwhile, according to NASA, the paired approach danger zone In the study, the position of the two aircraft in the pairing operation was determined by the longitudinal spacing. Therefore, this paper establishes a plane rectangular coordinate system and analyzes the problem in a two-dimensional plane motion model [14]. The relative coordinate system of the system are r, h and wi . They represent the linear distance between the two aircraft, the angle between the two lines from the x axis and the angle between the two lines connecting the counterclockwise rotation to the heading of aircraft i. The relative coordinate system of the avoidance aircraft corresponds to the position of the intrusion aircraft, respectively is xr , yr and br , and the vb direction of the yr axis is deﬁned by the formula (1)–(3). xr ¼ xe xb

ð1Þ

yr ¼ ye yb

ð2Þ

br ¼ be bb

ð3Þ

452

X. He et al.

Fig. 3. Relative motion reference system of two planes

The initial motion parameters for each aircraft are respectively vi , ui and wi þ dwi . They represent the speed of the aircraft i, the lateral movement speed and the turning rate. We let bbf indicate the heading of the aircraft, and the lower corner f indicate the termination status. The motion process of each aircraft in the reference frame is modeled and represented by Eqs. (4)–(6): x_ i ¼ vi cosðbi Þ þ ui sinðbi Þ

ð4Þ

y_ i ¼ vi sinðbi Þ ui cosðbi Þ

ð5Þ

b_ i ¼ xi þ dxi

ð6Þ

The range of input variables is represented by Eqs. (7)–(9): vmin vi vmax i i

ð7Þ

umax ui umax i i

ð8Þ

dxmax dxi dxmax i i

ð9Þ

For intrusion aircraft, it will usually be corrected in time due to the wrong approach, and no continuous turn will be made. Therefore, the turning rate xb is 0. Let dxb represent the change of the turning rate, and the range of the turning rate to the left and right is the same. For the avoidance aircraft, xe indicates the standard turning rate in the motorized avoidance procedure. Let dxe indicate the change in the turning rate. The relative motion model of the two aircraft is represented by Eqs. (10)–(12):

The Calculation of Safety Front Boundary of Paired Approach Procedure

453

r_ ¼ ve cosðwe Þ þ ue sinðwe Þ vb cosðwb Þ ub sinðwb Þ

ð10Þ

ve sinðwe Þ ue cosðwe Þ vb sinðwb Þ þ ub cosðwb Þ h_ ¼ r

ð11Þ

_ w_ i ¼ xi þ dxi h;

for i ¼ b; e

ð12Þ

In the calculation of the relative coordinate system, two different sets of states ðxr ; yr ; br Þ and ðr; wb ; we Þ will be used. The ﬁrst set of sets represents the intrusion area of the aircraft. The second set of sets is used to determine the initial motion parameters for the most dangerous situation. The initial motion parameters of each aircraft can be determined separately or based on the state of other aircraft. At present, the paired approach has been ofﬁcially implemented at the San Francisco International Airport in the United States, and the domestic research on paired approach is still in the theoretical stage. Therefore, the termination conditions calculated in this paper use the interval between the two aircraft speciﬁed by the FAA in the 8020.16 document. It’s less than 500 ft (153 m). Expressed by Eq. (13): T ¼ fr : r ¼ 500g

ð13Þ

Assume that the paired aircraft are near-miss. According to the initial control parameters obtained in this case, regardless of the accelerated escape of the avoidance aircraft, the flight path Eq. (2) is used to obtain the flight path of the aircraft, by formula (14) and (15). The parameter dt is expressed as tj þ 1 tj . xi ¼ xi þ 1 ½vj þ 1 cosðbj þ 1 Þ þ u sinðbj þ 1 Þdt

ð14Þ

yi ¼ yi þ 1 ½vj þ 1 sinðbj þ 1 Þ u cosðbj þ 1 Þdt

ð15Þ

For the case where the rear aircraft intrudes into the lead aircraft trajectory, this paper adopts the method of avoidance aircraft to accelerate the escape to avoid entering the aircraft to increase the safety margin; for the case where the lead aircraft breaks into the rear aircraft trajectory, the operation is difﬁcult and more urgent. This article only considers that the turning of the aircraft is less than 45°, avoiding the situation that the aircraft flees with no more than 60° to ensure the safety of the two aircraft. In order to facilitate the calculation of the minimum distance between the two aircraft in the collision avoidance maneuvers, the aircraft trajectory is obtained by continuously iterating and updating the aircraft position through time intervals, and the dichotomy is used to solve the problem. Based on the initial control parameters in this dangerous situation and the flight path equation of the aircraft, the minimum longitudinal safety interval between the two aircraft is calculated by MATLAB simulation, as shown in Fig. 4. In the case of avoidance aircraft’s 60-degree turn and fleeing, the two aircraft have similar motion parameters, the escaping angle of the avoidance aircraft is greater than the intrusion aircraft. At the same time, it is necessary to consider the response time of the avoidance aircraft, that is, the response time from the intrusion aircraft into the

454

X. He et al.

Fig. 4. Emulational illustration of minimum longitudinal separation of two planes

wrong approach to avoidance aircraft starting escape with a large angle turn. Since the pilot’s attention is very concentrated during the paired approach, the response time must not exceed 3 s.

4 Case Analysis The model uses two B747-400 s with the following parameters: the wingspan is 64.4 m, and the ﬁnal approach speed is 295 km/h. The left aircraft is the lead aircraft and the right aircraft is the trailing aircraft. When the lead aircraft erroneously turns to the course of the trailing aircraft during the ﬁnal approach phase, the trailing aircraft has a three-second response and establishes the slope time, then takes a larger angle to escape to avoid the danger, regardless of the acceleration at the time of escape. In the ﬁnal approach, after flying over the FAF, the two aircraft fly along the extension line of their respective runway center lines, and the lateral distance between the two aircraft is the runway spacing. Take Shanghai Hongqiao International Airport as an example. The centerline between two runways is 365 m apart. The extreme situation of avoidance is chosen: the angle of the intrusion into the aircraft is 45°, and the avoidance angle of the avoidance aircraft is 60°. The minimum spacing is 153 m (500 ft), which is the termination of the pursuit process. The minimum longitudinal safety interval between the two aircraft is 233 m by backlash of the flight path, as shown in Fig. 5(a), which determines the maneuver avoidance condition. The front boundary of the safety front boundary is not less than 233 m. This interval is a critical condition. If the longitudinal interval is larger than this value, the risk of collision between the two aircraft during the paired approach can be further reduced. On this basis, the turning angle of the intrusion into the aircraft is reduced to 30°, and the avoidance angle of the avoidance aircraft is reduced to 45°. In this case, the minimum longitudinal safety interval between the two aircraft determined by the flight path is 104 m. As shown in Fig. 5(b).

The Calculation of Safety Front Boundary of Paired Approach Procedure

455

Fig. 5. Chase illustration of 365 m of lateral spacing

5 Conclusion At present, operation speciﬁcations for China’s closely spaced parallel runways are still dominated by single-runway operation mode, and its advantages are not fully utilized. The paired approach mode can greatly improve the capacity of the closely spaced parallel runways. In this paper, the research on the paired approach in the case of maneuver avoidance is carried out, which further improves the feasibility of the application of the paired approach in the closely spaced parallel runways airport in China.

References 1. Stone, R.: Paired approach concept. In: Proceedings of the NASA Workshop on Flight Deck Centered Parallel Runway Approaches in IMC. NASA Langley Research Center, Hampton, VA, 29 October 1996 2. Hammer, J.: Case study of paired approach procedure to closely spaced parallel runway. Air Trafﬁc Control Q. 8, 223–252 (2000) 3. Bryant, T., Kuchar, J.K.: Evaluation of collision alerting system requirements for paired approach. In: 19th Digital Avionics Systems Conference (2000) 4. Teo, R., Tomlin, C.J.: Computing provably safe aircraft to aircraft spacing for closely spaced parallel approaches. In: Digital Avionics Systems Conference, vol. 1, 2D2/1–2D2/9 (2000) 5. Teo, R., Tomlin, C.J.: Computing danger zones for provably safe closely spaced parallel approaches. J. Guidance Control Dyn. 26(3), 434–442 (2003) 6. Teo, R., Tomlin, C.J.: Provably safe evasive maneuvers against blunders in closely spaced parallel approaches. In: AIAA Guidance, Navigation, & Control Conference, vol. 92, no. 1, pp. 173–180 (2013) 7. Eftekari, R.R., Hammer, J.B., Havens, D.A., et al.: Feasibility analyses for paired approach procedures for closely spaced parallel runways. In: Integrated Communications, Navigation and Surveillance Conference (ICNS), pp. 1–14 (2011)

456

X. He et al.

8. Johnson, S.C., Lohr, G.W., Mckissick, B.T., et al.: Simpliﬁed aircraft-based paired approach concept deﬁnition and initial analysis. Approach Control (2013) 9. Perry, R.B., Madden, M.M., Torrespomales, W., Butler, R.W.: The simpliﬁed aircraft-based paired approach with the alas alerting algorithm. Perry Raleigh B (2013) 10. Lu, F., Lv, Z.-P., Zhang, Z.-N., Wei, Z.-Q., Liu, B.-L.: longitudinal collision risk assessment on paired approach on closely spaced parallel runway. J. China Saf. Sci. 23(8), 108–113 (2013) 11. Niu, X.-L., Lv, Z.-P., Zhang, Z.-N.: The calculation of the minimum follow-up distance for paired approach for closely spaced parallel runway. Aeronaut. Comput. Tech. 2015(04), 46–48 (2015) 12. Tian, Y., Yan, Y.-J., Wan, L.-L., Ye, B.-J.: Internal analyses for paired approach for closely spaced parallel runway (05), 11–14 (2015) 13. Han, D.: Feasibility analyses for paired approach for closely spaced parallel runway in Shanghai Hongqiao Airport. Civil Aviation Flight University of China (2016) 14. Song, F.: Research on Safety boundary of Paired Approach Mode. Civil Aviation Flight University of China (2017) 15. Chen, Y., Han, D., He, X., et al.: The discussion of paired approach for closely spaced parallel runway. China Civ. Aviat. 5, 51–122 (2017, suspension of publication) 16. Xin, H., Jiang, H., Dan, H.: Safety boundary of paired approach for closely spaced parallel runway. Adv. Aeronaut. Sci. Eng. 8(3), 321–326 (2017) 17. Teo, R., Tomlin, C.J.: Computing danger zones for provably safe closely spaced parallel approaches. J. Guidance Control Dyn. 26, 434–442 (2003)

Research About Optimizing the Wake Turbulence Separation for Takeoff of CSPRs Under Crosswind Conditions Yaqing Chen1, Yujie Hou1(&), Dengfeng Hu2, and Chunzheng Wang2 1

2

CAAC Academy of Flight Technology and Safety, Civil Aviation Flight University of China, Guanghan, Sichuan 618307, China [email protected], [email protected] School of Air Trafﬁc Management, Civil Aviation Flight University of China, Guanghan 618307, China [email protected], [email protected]

Abstract. In order to increase the runway capacity of CRPRs, do researches on the decay and transport of wake vortex, then establish a model of wake turbulence separation for takeoff for closely spaced parallel runways. Comprehensively considering the factors affecting the takeoff wake interval model, and taking Shanghai Hongqiao International Airport as an example, this paper uses B747-400 as the lead aircraft and B737-700 as the trailing aircraft. It uses crosswind conditions to optimize the wake turbulence separation for takeoff by using MATLAB software and determines the favorable crosswind volume 2.5 m/s as the critical crosswind value of the takeoff interval without wake influence after optimization. The calculation results are compared with the value of Eurocontrol (European Organization for the Safety of Air Navigation) wake encounter risk, verifying that its safety performance fully meets actual operating requirements. Keywords: Closely spaced parallel runways Wake turbulence separation for takeoff Crosswind

1 Introduction In recent years, in the face of continued growth in air trafﬁc, increasing the number of airport runways can effectively increase airport capacity. The close parallel runway is widely used due to its small land occupation area and low construction cost. The wake interval is the main factor affecting runway capacity. During the take-off phase, CSPRs severely limit their capacity advantage due to the relatively conservative wake spacing. Domestic and foreign scholars have conducted a lot of research on the wake interval. J.N. Hallock and S.P. Osgood calculated the strength and dissipation time of Research supported by 2018 Civil Aviation Administration Civil Aviation Safety Capacity Building Funding Support Research Project (TM2018-3-1/2). © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 457–464, 2019. https://doi.org/10.1007/978-3-030-04582-1_53

458

Y. Chen et al.

the wake generated by different models of CSPRs at different ground speed [1] Frank Holzäpfel considered the factors of wind, turbulence, stability and ground effect, and proposed a two-stage wake stochastic dissipation model [2, 3], and used simulation software to evaluate the probability of takeoff departure wake at Frankfurt International Airport [4]. Fred H. Proctor and David W. Hamilton established a wake prediction model by measuring the aircraft wake at Denver International Airport in the United States [5]. Henrik Hesse and Rafael Palacios proposed a flexible aircraft time domain elastic simulation method that uses a closed-loop system to reflect the load on the aircraft’s control surface that encounters wakes [6]. Domestic studies on wake spacing have also been conducted. Hu Jun made a comparative analysis of the wake spacing criteria for typical applications at home and abroad, and established a wake spacing model [7]. Wei Zhiqiang established a vortex motion dissipation model based on foreign research [8]. Xu Xiaohao et al. applied a large eddy simulation method to numerically simulate the wake vortex flow ﬁeld of a 3D simpliﬁed wing model [9]. Xue Yuan used the corresponding flight risk probability values on the Copula model mesh nodes to construct a two-dimensional and three-dimensional visualization risk probability map in the wake ﬁeld [10]. At present, the research mainly focuses on the wake interval in the approach stage, and the research on the widely used CSPRs takeoff wake interval is lacking, and the lateral wind is not used as a favorable factor for the wake dissipation to optimize the existing interval standard. The author used to change the position and intensity of the aircraft wake and the wake resistance of the aircraft, and used the crosswind as the main factor for the dynamic adjustment of the wake interval. The wake drift model of the CSPRs take-off aircraft was constructed. The model achieves the research goal of optimizing the CSPRs takeoff wake interval based on crosswind. Finally, taking Shanghai Hongqiao Airport as an example, the short parallel runway wake interval is calculated, and the interval safety is evaluated according to the crosswind prediction ability and the pilot driving level. The calculation results show that under certain crosswind values, CSPRs can implement the takeoff interval value without wake influence, thus achieving the reduction optimization of the CSPRs takeoff wake interval. The research results can provide theoretical support for the calculation of CSPRs take-off wake interval, which is of great signiﬁcance for improving CSPRs capacity.

2 The Factors Affecting the Takeoff Wake Interval Model The speed of the wake dissipation is expressed as the rate of change of the wake intensity per unit time, and its intensity during the dissipation process is the main factor determining the aircraft interval. The vortex is the product of the lift obtained by the aircraft. According to the theorem of Kutta-Joukowski [5], the initial amount of air generated by the aircraft is: C0 ¼

W : qVb0

ð1Þ

Research About Optimizing the Wake Turbulence Separation

459

The vortex begins to enter the dissipation phase after formation through the roll-up zone. According to the dissipative empirical model and the LES experimental results, Frank obtained a two-stage stochastic dissipative model by ﬁtting, namely the initial diffusion phase and the rapid decay phase [5]. According to the research results at home and abroad, the motion of vortex can be divided into three stages, namely, the stage of remote movement, the stage of near-ground movement, and the stage of influence of ground effect [12]. In this paper, the primary part of the vortex is represented by the primary vortex, and the ground effect vortex is represented by the secondary vortex. Since the secondary vortex intensity is much smaller than the primary vortex intensity, only one vortex velocity is considered, According to the theory proposed by Liu [12], i ¼ 1; 2, j ¼ 1; 2; 3; 4. 8 P Cj zj zi C0i cos hi dyi > ¼ v þ > cw < dt 2p rij2 þ 2pr0 i j6¼i P Cj yj yi C0i sin hi : dz > i > : dt ¼ 2p rij2 þ 2pr0 j6¼i

ð2Þ

i

0

dhi Ci Ci ¼ : 0 dt 2p ri2

ð3Þ

2 2 rij2 ¼ yj yi þ zj zi :

ð4Þ

0

0

Ci is the secondary vortex intensity corresponding to the ﬁrst vortex of the No. i, rj is distance of the primary vortex and the secondary vortex, and c is the rolling angle of 0 0 them. ri ¼ b0 =2, Ci ¼ 0:4Ci , h0 ¼ 0:6b0 . It is necessary to consider the aircraft wake encounter degree and the wake motion dissipation model to determine the aircraft wake interval. This paper uses the rolling moment coefﬁcient to measure the ability of the aircraft to return to a safe state after encountering the wake. When the aircraft encounters a wake, the rolling moment coefﬁcient RMC can be expressed as a function of bbfl : Cv ARf bl F RMC ¼ Vf bf ARf þ 4 bf

ð5Þ

Cv is the aircraft strength from the front, where Vf is the flight speed, ARf is the rear wing span area, and bl ; bf are the front and rear wing extensions respectively. In addition, a ¼ 0:04. 0 1 sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ﬃ bl bl @ bl 2 bl A 2a ¼ 1 2 2a 1 þ 2a : F bf bf bf bf

ð6Þ

460

Y. Chen et al.

3 Model of Wake Interval of CSPRs Assume that the wake strength that the rear aircraft can withstand A. It can be obtained from the rolling moment coefﬁcient in the encounter model that: Ccrit

RMCcrit ARf bl F ¼ : Vf bf ARf þ 4 bf

ð7Þ

Where RMCcrit is the critical roll factor. It is known that the wake produces a height, and the time for the wake to dissipate to Ccrit by the dissipation model is tcrit . 0 tcrit ¼tcrit t tcrit

¼

Ccrit1

Ccrit C0

ð8Þ ð9Þ

1

Ccrit is the inverse function of Ccrit . Deﬁne ts as the current wake interval standard Tmax ¼ max½ts ; tcrit , Tmin ¼ min½ts ; tcrit As shown in Fig. 1, there are two CSPRs. The left side is the front takeoff runway (RWY L), and the right side is the rear takeoff runway (RWY R). The left side of the wind is unfavorable crosswind. Negative values represent the right side of the wind and are favorable for crosswinds.

Fig. 1. Runways of aircraft takeoff

In order to ensure the safety of the take-off takeoff, the safety margin is set in this study. Let the safety margin be d. It is considered dangerous when the wake intensity is greater than the critical wake strength and the left boundary of the right runway is less than d.

Research About Optimizing the Wake Turbulence Separation

461

4 Case Analysis Shanghai Hongqiao Airport has two runways: 36L/18R and 36R/18L. These are typical closely spaced parallel runways. According to the statistics of flight take-off and landing in a certain week in June 2017, the B747-400 was selected as the front aircraft, and the B737-700 was used as the rear aircraft to calculate the typical model. The parameters are shown in the Table 1 below. Table 1. The data of front and rear aircraft model Aircraft type

B747400 B737700

Parameter(unit) Wingspan Take-off (m) distance (m)

Maximum take-off weight (kg)

Maximum landing weight (kg)

Maximum climb rate(m/s)

Limit of wind speed(m/s) 90 degrees crosswind

Following wind

64.4

3200

396890

285766

10.2

15

7.7

34.31

2316

70553

65310

22.7

18

5

According to the wake encounter model and the parameters of the aircraft, the critical strength of the wake that the B737-700 can withstand can be calculated as Ccrit ¼ 200m2 =s. The existing wake turbulence separation for takeoff is 2 min [13]. Tmax ¼ 120s. The safety margin is taken as d ¼ 50m. The resulting security zone boundary is S ¼ 292:5m. (1) when crosswind volume is 0, vcw ¼ 0. The trajectory of the vortex when the front wheel is lifted at Tmax ¼ 120s is shown in Fig. 2. In the absence of crosswind, The two vortices move farthest to the y0 ¼ 206m at Tmax on the inside and move up to 73 m. Wake flow trajectory at 1:5b0 without crosswind is shown in Fig. 3. In the absence of crosswind, the vortex move farthest to the y1:5b0 ¼ 129m at 1:5b0 .

Fig. 2. The trajectory of the vortex when the front wheel is lifted without crosswind

462

Y. Chen et al.

Fig. 3. Wake flow trajectory at 1:5b0 without crosswind

(2) when crosswind volume is 0.5 m/s,vcw ¼ 0:5 m=s The trajectory of the vortex when the front wheel is lifted at Tmax ¼ 120s is shown in Fig. 4. Wake flow trajectory at 1:5b0 without crosswind is shown in Fig. 5. Compared with the bounce trajectory without crosswind, The vortex move farthest to the y1:5b0 ¼ 165m at 1:5b0 on the inside.

Fig. 4. The trajectory of the vortex when the front wheel is lifted with crosswind at 0.5 m/s

Fig. 5. Wake flow trajectory at 1:5b0 with crosswind at 0.5 m/s

Research About Optimizing the Wake Turbulence Separation

463

The trajectory of vortex can be obtained from the two cases. ymax ¼ 265m. Under the condition of crosswind at 0:5m=s, the wake generated by the ﬁrst takeoff aircraft of the 36L/18R runway will not affect the takeoff aircraft after the 36L/18R runway. (3) when crosswind volume is 1 m/s, vcw ¼ 1m=s The trajectory of the vortex generated in the 120 s when the front wheel is lifted is as shown in Fig. 6. It can be concluded that the wake generated by the ﬁrst takeoff aircraft of 36L/18R runway will affect the takeoff aircraft after the 36R/18L runway.

Fig. 6. The trajectory of the vortex when the front wheel is lifted with crosswind at 1 m/s

The takeoff wake interval under crosswind optimization mainly depends on the influence of crosswind on the wake dissipation. According to the current domestic wind speed prediction level [14–16], the relatively conservative wind speed prediction parameters are as follows: lw ¼ 0:7, rw ¼ 0:64. Based on the existing statistical results [17], the relevant parameters are as follows:lf ¼ 4, rf ¼ 15. The wake encountering risk under crosswind optimization is Po ¼ 0:003. According to the research results of the CREDOS project, the risk encountered under the current takeoff wake interval under windless conditions is taken as the critical wake encounter risk value [18]. Pc ¼ 0:004. The comparison can lead to conclusions that Po \Pc . Therefore, the use of crosswind-optimized Hongqiao Airport takeoff wake interval values meets existing safety performance requirements, and the results are feasible.

5 Conclusion Taking Shanghai Hongqiao Airport as an example, the results show that when the favorable crosswind volume is 2.5 m/s, the lead aircraft is B747-400, and the trailing aircraft B737-700 can be equipped with take-off interval without wake influence. In the follow-up study, it is necessary to improve the cross-wind forecasting ability, in order to obtain the implementation conditions with sufﬁcient take-off interval without wake.

464

Y. Chen et al.

References 1. Hallock, J.N., Osgood, S.P.: Wake vortex effects on parallel runway operations. In: 41st Aerospace Sciences Meeting and Exhibit, 6–9, Reno, Nevada: AIAA 2003, p. 379 (2003) 2. Holzäpfel, F.: Probabilistic two-phase wake vortex decay and transport model. J. Aircr. 40 (2), 323–331 (2003) 3. Holzäpfel, F.: Probabilistic two-phase aircraft wake-vortex model: further development and assessment. J. Aircr. 43(3), 700–708 (2006) 4. Holzäpfel, F., Kladetzke, J.: Assessment of wake-vortex encounter probabilities for crosswind departure scenarios. J. Aircr. 48(3), 812–822 (2011) 5. Proctor, F.H., Hamilton, D.W.: Evaluation of fast-time wake vortex prediction models. In: 47th AIAA Aerospace Sciences Meeting Including The New Horizons Forum and Aerospace Exposition, Orlando, Florida: AIAA 2009, p. 344 (2009) 6. Jun, H.: Study on the safety interval of wake in air trafﬁc. Nanjing University of Aeronautics and Astronautics, Nanjing (2001) 7. Zhiqiang, W.: The Research Modeling and Simulation of Flow Field and Safety Spacing for Wake Vortex. Civil Aviation University of China, Tianjing (2008) 8. Xiaohao, X.: Large eddy simulation of wake vortex during approach. J. Nanjing Univ. Aeronaut. Astronaut. 42(2), 179–184 (2010) 9. Winckelmans, G., Duquesne, T., et al.: Summary Description of the Models Used in the Vortex Forecast System (VFS). Catholic University of Louvain, Belgium (2005) 10. Sarpkaya, T.: New model for vortex decay in the atmosphere. J. Aircr. 37(1), 53–61 (2000) 11. Robins, R.E., Delisi, D.P., Greene, G.C.: Algorithm for prediction of trailing vortex evolution. J. Aircr. 38(5), 911–917 (2001) 12. Liu, H.T.: Tow-tank simulation of vortex wake dynamics. In: FAA, Proceedings of the Aircraft Wake Vortices Conference, 29–31 October 1991 13. CCAR-93TM-R5: Air trafﬁc management rules for civil aviation of China. Beijing: Civil Aviation Administration of China (2017) 14. Research on Wind Speed Forecasting Based on Error Correction and Fuzzy Evaluation, Taiyuan University of Technology, Taiyuan (2016) 15. Study on Very Short Term Wind Speed and Short Term Wind Power Prediction Based on Error Analysis, Shandong University, Jinan (2016) 16. Research on Prediction Method of Ultra-Short-Term Wind Speed of Wind Farm Based on Normal Distribution Noise Neural Network, Lanzhou University of Technology, Lanzhou (2017) 17. Hall, T., Soares, M.: Analysis of localizer and glide slope flight technical error. In: Digital Avionics Systems Conference, 2.D.2-1-2.D.2-9 (2008) 18. Kauertz, S., Holzäpfel, F., Kladetzke, J.: Wake vortex encounter risk assessment for crosswind departures. J. Aircr. 49(1), 281–291 (2012)

A Generalized FSM Implementation Framework Mao-Hsiung Hung1,2(&) 1

College of Information Science and Engineering, Fujian University of Technology, No. 33, Xuefunan Road, University Town, Minhou, Fuzhou 350118, China [email protected] 2 Fujian Provincial Key Laboratory of Big Data Mining and Applications, Fujian University of Technology, No. 33, Xuefunan Road, University Town, Minhou, Fuzhou 350118, China

Abstract. Finite State Machine (FSM) is a useful and powerful tool to model a dynamic system. On the traditional implementation of FSM, nested switch-case statement exists several problems of poor reusability and maintainability. The paper presents a generalized framework of algorithmic implementation to improve these disadvantages of nested switch-case statement. A lookup table and event handle is used to construct various FSMs in the proposed framework. In the experiments, we apply the proposed framework to practice a typical FSM. The experimental results demonstrates the framework’s generalization. Keywords: FSM

Implementation framework Generalized design

1 Introduction Finite State Machine (FSM) provides a useful and powerful tool to model a dynamic system. Particularly, FSMs can effectively and efﬁciently describe complex logics of systems, so that engineers and programmers apply FSM to avoid designing heuristically for complex systems. Therefore, FSMs have receiving a largely wide applications and developments in many engineering and scientiﬁc domains. A traditional and simple programming implement uses nested switch-case statement to construct FSMs [1]. However, the implement framework of nested switch-case statement has poor reusability and its codes are difﬁcult to maintain because of simple data structure. Xu et al. [2] use a object-oriented programming to encapsulate the nested switch-case statement and enhance FSM’s data structure, so that the reusability and maintainability of the framework is obtained improvement. However, the core part of the nested switch-case statement is still needed rewriting to adapt various FSMs in the object-oriented framework. Other object-oriented design is proposed in [3]. Several works developed drawing and visualization tools to plot and describe state transition diagrams of FSMs such as Finite State Machine Editor [4] and Matlab Stateflow toolbox [7], and then these tools can automatically generate FSM codes. Niaz et al. [8] describes an object-oriented (OO) approach to generate compact and efﬁcient Java code from UML statechart diagrams. However, when the modiﬁcation of the state diagrams © Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 465–470, 2019. https://doi.org/10.1007/978-3-030-04582-1_54

466

M.-H. Hung

are demanded, these FSM codes are again generated and deployed to the implement framework. Juhasz et al. propose an active library to implement FSMs using template metaprogramming [5]. In this work, we developed an implementation framework for FSMs, featured by high generalization. A lookup table is applied to represent a state transition of a FSM in our proposed framework. Cooperating with the lookup table, we propose event handle algorithms to perform state transition and output action of FSMs. The proposed implementation framework achieves that the conﬁguration of the lookup table is only required to construct various FSMs without changing the core codes. In the experiments, we practiced a typical FSM to demonstrate the generalization of the proposed framework. That is promising to bring beneﬁts for hardware design of FSMs [6].

2 Proposed Method To represent a state transition diagram of a FSM, we apply a lookup table to store the conﬁguration of all state transitions of the FSM. Each row of the lookup table is used to represent a state transition. We designed a data structure in a row, which contains four ﬁelds to record a transition. The ﬁve ﬁelds are denoted by fromState, event, toState and act. The fromState ﬁeld records the starting state of a transition. The event ﬁeld records a speciﬁc event to trig the state transition. The toState ﬁeld records the ending state of the transition. The act ﬁeld stores a runner of an action which is performed once the state transition happens. Figure 1(a) shows a single state transition from a state of X to a state of Y driven by a event of I to perform a action of O. Table 1 is a lookup table of Tab and its rows of Tab[0], Tab[1], …, Tab[N–1] are used to represent N state transitions of a FSM.

X

I/O

Y

Fig. 1. Single transition without condition

Table 1. Data structure of lookup table fromState event toState act Tab[0] Tab[1] … Tab[N–1]

After the construction of the lookup table of Tab, the event handle of the FSM are required to proposed. We separate into two algorithms to describe the two event handles without condition and with condition. Algorithm I is proposed to process the

A Generalized FSM Implementation Framework

467

event handle without condition which is written by a function of EventHandle(e) inputting a event of e as follows. Algorithm I. Event handle of FSM without condition Line 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Pseudo code Input: a event of e function EventHandle(e) cs←curStateFsm, ns←cs, act←null, flag←false for i←0 to N−1 do if cs=Tab[i].fromState and e=Tab[i].event then ns←Tab[i].toState, act←Tab[i].act, flag←true exit for-loop end if end for if flag then if act≠null then call act.run() else do nothing curStateFsm←ns else do nothing end if

In the algorithm, we ﬁrst deﬁne a variable of curStateFsm to store the current state of the FSM. The beginning of the FSM execution initializes curStateFsm variable to one of state. Then, one of events happens to trig a state transition and an action performance. In Line 3, four local variables of cs, ns, act and flag are initialized. cs and ns are assigned by curStateFsm and act is assigned by null. The null value makes the variable not to refer any object. flag is assigned by false that means no matching for the current state and the event input before the table lookups. After the variable initialization, we try to travel all rows of the lookup table of Tab using a for-loop in Line 4–9. And then, we match cs and e with Tab[i].fromState and Tab[i].event for i = 0,1, …, N–1 in Line 5. In Line 6, once the match of cs and e hits, ns is assigned by Tab[i].toState and act is assigned by Tab[i].act. As a result, the next state of the trigged state transition is ready in ns and act refers to the corresponding action runner. The flag changes to true that means a successful matching for the current state and the event input, and then we exits the for-loop. If cs and e have no matching with Tab[0], Tab[1], …, Tab[N-1], flag will keep false. After the matching of cs and e, we check whether flag is true in Line 11. If yes, then we check whether act6¼null. If act6¼null, then an action is required to execute between the state transition, so we call act.run(), otherwise nothing performs, as the described in Line 12–14. Then in Line 15, curStateFsm is assigned by ns i.e. the next state, and the state transition ﬁnishes. If flag6¼true, that means any state transition and any action will not perform.

468

M.-H. Hung

3 Experimental Result In our proposed method, the lookup table structure and the event handle function are the core parts of FSM framework. The core part is basically unchanged. We ﬁrst deﬁne symbols of states and action, and then conﬁgure the lookup table to build various of FSMs, so that it makes the framework to achieve generalized purposes. To demonstrate the generalized ability, we applied our proposed framework to a typical FSM of elevator door controlling. The FSM of elevator door controlling has four states of the door including opened, closing, closed and opening. The two events for door’s button are “request to open” and “request to close”. Another two events for door’s sensor are “sensor closed” and “sensor opened”. The three actions of the door are “move to close”, “move to open”, and “stop moving”. We enumerated symbols to represent states, events and actions before the construction of the state transitions of FSMs. Table 2 lists the meanings and symbols of elevator door controlling FSM. These symbols include states of S1, S2, S3 and S4, events of I1, I2, I3 and I4, and actions of O1, O2 and O3. Based on these symbols, the state transition diagram of the FSM is as shown in Fig. 2.

Table 2. Meaning and symbol of FSM of elevator door controlling Meaning Door is opened Door is closing Door is closed Door is opening Event Request to close Sensor closed Request to open Sensor opened Action Door moves to close Door moves to open Door stops moving State

I1/O1

S1

I4/O3

Symbol S1 S2 S3 S4 I1 I2 I3 I4 O1 O2 O3

I3/O2 S4

S2 I1/O1

I3/O2

S3

I2/O3

Fig. 2. State transition diagram of elevator door controlling

A Generalized FSM Implementation Framework

469

According to the state transition diagram, we conﬁgure the lookup table as listed in Table 3. The six rows of Tab[0], Tab[1], …, Tab[5] represent the six transition in the diagram. For an example, the transition of I1/O1 from S1 to S2 is assigned by S1, I1, S2 and ActO1 respectively in fromState, event, toState and act ﬁelds of Tab[0]. Recorded in the act ﬁeld, ActO1, ActO2 and ActO3 mean action runners respectively corresponding to O1, O2 and O3. Each action runner contains a function of run() to perform a speciﬁc action. For an example, the action runner of ActO1 contains ActO1. run() function to perform O1 action. Table 3. Conﬁguration of Lookup table of FSM of elevator door controlling Tab[0] Tab[1] Tab[2] Tab[3] Tab[4] Tab[5]

fromState S1 S2 S3 S4 S2 S4

event I1 I2 I3 I4 I3 I1

toState S2 S3 S4 S1 S4 S2

act ActO1 ActO3 ActO2 ActO3 ActO2 ActO1

During simulation phrase of the FSM, we ﬁrst assigned the initial state to S1, i.e. curStateFSM←S1, and then inputted an event sequence of {I1, I2, I3, I4, I3, I1} and tested the function of EventHandle(e) in one event at a time. The testing outputs displayed that the FSM transferred state by a sequence of {S1, S2, S3, S4, S2, S4, S1} and performed a series of actions of {ActO1, ActO3, ActO2, ActO1, ActO2, ActO3}, as shown in Table 4. The demonstration unveils that no changing is required in algorithm of EventHandle(e) and data structure of the lookup table. As a result, good generalization is archived by the proposed framework. Table 4. Simulation output of elevator door controlling FSM Line 1 2 3 4 5 6 7 8 9 10 11

Output Start to test an elevator door FSM. Current state is S1(Opened) I1(Request to close) is coming. ActO1: Door moves to close. Current state is S2(Closing) I2(Sensor closed) is coming. ActO3: Door stops moving. Current state is S3 Closed) I3(Request to open) is coming. ActO2: Door moves to open. Current state is S4(Opening) (continued)

470

M.-H. Hung Table 4. (continued) Line 12 13 14 15 16 17 18 19 20 21

Output I1(Request to close) is coming. ActO1: Door moves to close. Current state is S2(Closing) I3(Request to open) is coming. ActO2: Door moves to open. Current state is S4(Opening) I4(Sensor opened) is coming. ActO3: Door stops moving. Current state is S1(Opened) Finish testing.

4 Conclusion This paper has presented a generalized framework on algorithmic implementation for FSMs. Dou to poor generalization, the traditional implementations of nested switchcase statement is required to modiﬁcation for different FSM applications. Cooperating with different lookup tables, our proposed framework can be applied to various FSMs without changes. The experimental results indicate that good generalization is achieved by the proposed framework. Moreover, the proposed algorithmic implementation will become good reference on hardware design of FSMs. Acknowledgement. This work was supported in part by Fujian Provincial Department of Science and Technology, Granted No. 2017J01729.

References 1. van Gurp, J., Bosch, J.: On the implementation of ﬁnite state machines. In: 3rd Annual IASTED International Conference on Software Engineering and Applications, vol. 3, no. 1, pp. 1–15 (1999) 2. Xu, X.-l., Wang, L.-y., Zhou, H.: Implementation framework of ﬁnite state machines. J. Eng. Des. 10(5), 251–255 (2003). (In Chinese) 3. Ackroyd, M.: Object-oriented design of a ﬁnite state machine. J. Object Oriented Program. (1995) 4. Finite State Machine Editor. http://fsme.sourceforge.net/ 5. Juhász, Z., Sipos, Á., Porkoláb, Z.: Implementation of a ﬁnite state machine with active libraries in C ++. In: Generative and Transformational Techniques in Software Engineering II, International Summer School, GTTSE 2007, Braga, Portugal, 2–7 July 2007. Revised Papers DBLP, pp. 474–488 (2007) 6. Cheng-Juei, Yu., Yi-Hsin, W., Wang, S.-D.: An approach to the design of speciﬁc hardware circuits from C programs. J. Inf. Sci. Eng. 34, 337–351 (2018) 7. Mass. Natick, Stateflow User’s Guide, The MathWorks, Inc. (2002) 8. Niaz, I.A., Tanaka, J.: An object-oriented approach to generate java code from UML statecharts. In: International Symposium on Computer and Information Sciences (2005)

Method on Scale Problem of Passenger Catering Center for High-Speed Railway Geng Jingchun(&) China Railway Design Corporation, Tianjin 300251, China [email protected]

Abstract. With the construction and operation of high-speed railway and intercity railway, Electric Multiple Units (EMU) passenger trains have been widely used nationwide, which play increasingly important roles in the railway passenger transport market; to meet the diversity requirements of EMU passengers, this paper discusses the types and distribution methods of breakfast, dinner, snack foods, etc. Firstly, catering principles and constraints of passenger catering center were analyzed, and then an optimized model was put forward for the scale problem of passenger catering center for high-speed railway, ﬁnally an example for the catering center scale of Tianjin railway hub was calculated, the result shows that the model could solve the catering center problem reasonably. Keywords: High-speed railway

Catering center Scale Passenger

On May 15, 2016, a new train diagram was implemented on China Railway Corporation, and great changes have taken place on the train operation plan in the past 10 years, 3,400 pairs of passenger trains are operated every day, among them, there are more than 2,100 pairs of EMU passenger trains, which account for 43.7% of the total passenger trafﬁc; with the construction and operation of high-speed railway and intercity railway, EMU passenger trains will be widely used nationwide, which will play increasingly important roles in the railway passenger transport market. However, due to the construction of passenger train vehicles and other factors, it is impossible to process food raw materials on the dining car like the passenger trains pulled by the generalspeed locomotives and provide ordinary meals to the passengers. With the development of the national economy, people’s requirements for the quality and level of material life will be higher and higher. How to solve and provide the EMU passenger train meals, which is one of the important factors that restrict the railway improving service quality. Meal allocation methods of the domestic and international airline company was fully summarized, for EMU passenger trains lots of advanced, standardized, large-scaled, high-quality, environmental-friendly and efﬁcient railway catering centers should be built, which could improve train services level; but a catering center cost much, so the scale of catering center has become an urgent problem to be solved.

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 471–476, 2019. https://doi.org/10.1007/978-3-030-04582-1_55

472

G. Jingchun

1 Meal Items and Distribution Methods for Catering Center 1.1

Meal Items Design

The EMU passenger trains have the advantages of fast running speed, high comfort and strong timeliness. For EMU passenger trains, the passenger consuming ability is generally between the airline passenger and the general-speed train passenger, the passenger flow was mainly included by business, commuting, and student. Generally, it has a certain economic foundation, which not only requires high quality of meals, but it also places higher demands on the items of meals. With reference to the items of meals served by the aviation catering, taking into account the way in which the meals on EMU passenger trains are sold, meals like breakfast, dinner, snack food, etc. should be provided. (1) Dinner, breakfast, fruit plates, and special meals for very important persons; (2) Dinner, breakfast, and fruit plates are mainly Chinese food, taking into account Western fast food, and should consider the special needs of the Muslim passengers; (3) The dinner is “rice, two meat dishes and one vegetable dish” and “rice, two vegetable dishes and one meat dish”, mainly based on hot dishes, while considering the soups that can be prepared; (4) The breakfast is mainly available in Chinese and Western styles, including fritters, bread, milk, soy milk, coffee, etc. (5) Special meals: customized meals for very important persons and so on; (6) Snack food: dessert, fruit plate, ice cream, hot and cold drinks, snack, dried fruit, small packaged wine, etc. 1.2

Meal Distribution Method

According to railway passengers’ large demand for meals, the catering center is generally far away from the hub passenger station. It can use two-stage distribution method, which is constituted with catering center and hub passenger transfer station. At ﬁrst, after the catering center has processed the meal, it will be stored in the cold storage; then, according to the food demand situation of each passenger station in the hub, the food will be transferred by the refrigerated container truck from the catering center to the passenger transfer station; ﬁnally, according to the conﬁguration requirements of the passenger trains of the EMUs, the meals will be transferred by insulation cars from the passenger transfer station to the freezer of the EMU train. The whole distribution process of the meal adopts the cold chain distribution method to ensure that the temperature of the meals is within the whole controllable range. The distribution process of the meal is shown in Fig. 1.

2 Scale Model for the Passenger Catering Center Whether the scale of the high-speed railway passenger catering center is reasonable, it directly affects the quality and level of railway service. The large scale of the catering center will result in waste of engineering investment, idle equipment, overcapacity,

Method on Scale Problem of Passenger Catering Center for High-Speed Railway

473

cold storage

cold storage

cold storage

catering center

hub passenger transfer station

EMU train

refrigerated container truck

Insulation car

Fig. 1. Schematic diagram of distribution method of meals

etc.; the small scale can’t meet the passengers’ dining needs, which directly affects the improvement of railway service quality. At present, although the relevant railway engineering design technical manuals and research reports [1, 2] are slightly related to the catering center, they have not formed a relevant theory to determine the scale of the catering center. This paper explores the applicable model of the scale of the high-speed railway catering center in China. 2.1

Catering Principles and Constraints

(1) According to the requirements of specialization, intensiﬁcation, and economics, combined with the number of EMU passenger trains delivered at the railway hub or region. Generally, only one catering center is set up in a railway hub or region to undertake the catering task of all EMU passenger trains; (2) In combination with the running distance and running time of the EMU passenger train, only the EMU passenger trains that cross the meal time during the operation period will be provided with meals; Deﬁne Li as the running distance of train i, Vi as the average travel speed of train i, Ti as the in-transit running time of train i; the decision variable Xi indicates whether the train needs to provide meals for passengers.

0 Li \ 700; Ti \ 3 1 or Ti ¼ Li =Vi

Xi ¼

(3) Each passenger train only provides meals at the starting and ﬁnishing station, and it will not provide supplementary meals at the halfway stop; for the trains that cross the meal time multiple times during the in-transit operation period, they are conﬁgured separately according to the times of having meals. The times of having meals during the in-transit operation period of train i is deﬁned as Mi , which is the times provided with meals for passengers.

474

G. Jingchun

Mi ¼

1 2

Li \ 700; Ti 3 or 700 Li 2000; 4 Ti 8 Li [ 2000; Ti [ 8

(4) Each passenger train is conﬁgured with catering, the amount of which is determined by the number of carriages, train standard capacity, passenger attendance rate and the rate of passenger dining. Deﬁne Pi as the standard capacity of train i, bi as the average attendance rate of passengers on train i, ci as the average rate of passenger dining on train i, and Ni as the amount of meals required for train i. Ni ¼ Pi bi ci vi Mi

2.2

Optimized Solution of the Catering Scale Model

There is usually only one catering center responsible for all EMU passenger trains in some particular railway hubs. Catering scale (Num) of catering center is decided according to the starting and ﬁnishing station, running distance, running time, train capacity, passenger attendance rate, and the rate of passenger dining of the passenger train operation plan. Num ¼

N X i¼1

Ni ¼

N X

Pi bi ci vi Mi

i¼1

3 Case Discussion Take Tianjin Railway Hub as an example to study the scale of catering center; Tianjin Railway Hub is located in Tianjin, north of North China. It is the intersection of Jinshan and Beijing-Shanghai railway arteries. It is the starting point of Beijing-Tianjin Intercity Railway, Tianjin-Qinhuangdao Passenger-dedicated Railway and Tianjin-Baoding Railway, Tianjin-Bazhou Railway, Tianjin-Jizhou Railway, Tianjin-Shanhaiguan Railway and Tianjin-Huanghua Railway, which is connected to Beijing, Shanghai, Shanhaiguan, Bazhou, Jixian, Huangqi and other six directions. It is also the throat inside and outside, an important trafﬁc artery in the northeast, north China and eastern China, a major railway hub with passengers and cargo and large railway hub of road and port intermodal transportation. Tianjin Railway Hub plays an important role in connecting and supporting Tianjin’s national economic development and even national transportation. With the passenger transportation systems of Beijing-Shanghai Highspeed Railway, Tianjin-Qinhuangdao Passenger-dedicated Railway, Tianjin-Baoding Railway, Beijing-Tianjin Intercity Railway and its extension lines being put into operation, Tianjin Railway Hub has ﬁnally formed the passenger transportation layout

Method on Scale Problem of Passenger Catering Center for High-Speed Railway

475

of “four main stations and seven auxiliary stations”. Four main stations are composed of Tianjin Railway Station, West Tianjin Railway Station, Binhai Station and Yujiapu Station, while seven auxiliary stations are composed of Tianjin South Railway Station, North Binhai Railway Station, Tianjin North Railway Station, Airport Station, North Junliangcheng Railway Station, Wuqing Railway Station and Tanggu Station. Caozhuang EMU Depot is responsible for planning and distributing the catering assignment of all originate and terminal trains, based on the passenger flow forecast and the passenger train operation plan, as shown in Table 1. Table 1. Train number serviced by catering center (Unit: pair/day) Year Train distance Tianjin Railway Station West Tianjin Railway Station Total

2020 700–2000 km 39 57

Above 2000 km 2 2

96

4

2030 700–2000 km 55 82 137

Above 2000 km 3 1 4

Based on the model of catering scale above, take the passenger train operation plan of Tianjin West Railway Station to Shenyang North Railway Station as an example, the amount of meals is calculated as follows: Originated from Tianjin West Railway Station to stop at Shenyang North Railway Station, passing through Tianjin-Qinhuangdao Passenger-dedicated Railway (300 km/h) and Qinhuangdao-Shenyang Passenger-dedicated Railway, trains are marshalling 8 carriages with a standard capacity of Pi ¼ 1200 people, a passenger attendance rate of bi ¼ 85% and a passenger dining rate of ci ¼ 30%, whose operation distance Li ¼ 750 km and operation time Ti ¼ 4:05 h. Thus it can be known that EMU passenger trains from Tianjin West Railway Station to Shenyang North Railway Station meet the condition of 700 Li 2000 and 4 Ti 8, which lead to Mi ¼ 1; vi ¼ 1. That is: Ni ¼ 1200 85% 30% 1 1 ¼ 306 Packages From this we can get that each train (marshalling 8 carriages) from Tianjin West Railway Station to Shenyang North Railway Station is in need of 153 packages of meals. Based on the model above, EMU trains in Table 1 are classiﬁed according to the original station, terminal station, operation path, operation speed and operation time of the passenger train operation plan. Take 2030 as an example, the number of trains who run a distance between 700 km and 2000 km in between 4 h and 8 h, also known as i7002000 , is 137, while the number of trains who run a distance above 2000 km in above 8 h, also known as i2000 , is 4. i7002000 ¼ 55 þ 82 ¼ 137 Trains i2000 ¼ 3 þ 1 ¼ 4 Trains

476

G. Jingchun

On the basis of train marshalling 16 carriages, standard capacity as 1200 people, passenger attendance rate as 85% and a passenger dining rate as 30%, the calculation of the packages of catering of Tianjin West Railway Station and Tianjin Railway Station is as follows: Num ¼ Num7002000 þ N2000 ¼ ði7002000 Mi vi þ i2000 Mi vi Þ Pi bi ci ¼ ð137 1 1 þ 4 2 1Þ 1200 85% 30% ¼ 44370 As a result, Caozhuang EMU Depot catering center should prepare 44370 packages of catering every day in 2030. Similarly, 31824 packages of catering are needed every day in 2020, as shown in Table 2. Table 2. Amount of meals serviced by catering center (Unit: package/day) Year Tianjin Railway Station Tianjin West Railway Station Total

2020 13158 18666 31824

2030 18666 25704 44370

4 Summary Through the analysis of the items of catering provided in high-speed railway transportation and products distribution methods, various kinds of safe food are processed and packaged together to provide a high-quality and diversiﬁed catering service and choice for passengers according to scientiﬁc and reasonable dietary method. Combined with the analysis of Caozhuang EMU Depot catering center of Tianjin Railway Hub, on the basis of catering principles, constraint condition and catering scale calculation model, the catering scale of Caozhuang EMU Depot catering center is checked to be 21 thousand packages per day. This also provides foundation for determining the scale of high-speed railway catering center.

References 1. The Second Survey and Design Institute of the Ministry of railways, the Third Survey and Design Institute of the Ministry of railways. The manual of railway engineering design technology, railway volume and organization, China Railway Press, Beijing (1992) 2. Railway Third Survey and Design Institute Group Co., Ltd. The feasibility study report of the project construction of the Beijing motor train base supporting project. Railway Survey and Design Institute Group Co., Ltd. (2009) 3. Railway Third Survey and Design Institute Group Co., Ltd. The feasibility study report on the distribution center, driving apartment and washing center of Cao Zhuang’s motor train section. Railway Survey and Design Institute Group Co., Ltd. (2012) 4. TB 10621-2014, design speciﬁcation for high speed railway. China Railway Publishing House, Beijing (2014)

An Improved Linear Population Size Reduction Based Parameters with Adaptive Learning Mechanism Diﬀerential Evolution (iLPALMDE) for Real-Parameter Single Objective Black Box Optimization Zhenyu Meng1 , Jeng-Shyang Pan1(B) , Wei-min Zheng2 , and Xiaoqing Li1 1

Fujian Provincial Key Lab of Big Data Mining and Applications, Fujian University of Technology, Fuzhou, China [email protected], [email protected], [email protected] 2 Shandong University of Science and Technology, Qingdao, China [email protected]

Abstract. In this paper, we proposed an improved Linear population size reduction based Parameters with Adaptive Learning Mechanism Diﬀerential Evolution (iLPALMDE) for real-parameter single objective black box optimization. The LPALMDE algorithm is a state-of-the-art DE variant proposed recently, nevertheless, it still has some weakness, e.g. the update scheme of crossover rate is heavily dependent on the number of individuals in each group, which may fall into a bad adaptation of crossover rate when population size becomes small in the linear population size reduction scheme. Therefore, a new adaptation scheme of crossover rate was advanced in this paper to tackle the weakness. This novel improved algorithm is veriﬁed under the CEC2013 benchmarks, and experiment results shows that it was competitive with other state of the art DE variants. Keywords: Benchmark functions Linear reduction · PALM

1

· Diﬀerential evolution

Introduction

There are many evolutionary algorithms proposed nowadays, e.g. Particle Swarm Optimization [1] (PSO), Diﬀerential Evolution [2–5] (DE), Ebb-Tide-Fish (ETF) algorithm [6,7], Monkey King Evolution (MKE) algorithm [8,9], QUATRE algorithm [10–14], etc., to tackle complex real-world optimization problems. By reviewing recent competitions, we can see that DE variants were welcomed by researchers and some recently proposed state-of-the-art DE variants secured ﬁrst ranks of these competitions. DE algorithm originated with Genetic Annealing c Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 477–484, 2019. https://doi.org/10.1007/978-3-030-04582-1_56

478

Z. Meng et al.

algorithm [3], a combined Genetic Algorithm [15] and Simulated Annealing [16], therefore, mutation operation, crossover operation and selection operation were inherited into DE as well. Accordingly, there are three control parameters involved into these three operations, F denotes scale factor which restricts the diﬀerence vector pair, Cr denotes the crossover rate which determines how many parameters of the donor vector inherited into the trial vector, ps denotes the population size which also deﬁnes number of selection operations each generation. The recommend settings of these parameters are given as follows: F = 0.5, Cr = 0.1 or Cr = 0.9, ps = 100 for diﬀerent optimization problems, e.g. DE/best/1 mutation strategy usually employed a small crossover rate Cr = 0.1 for optimization problems and it performed very well on unimodal or separable objectives while DE/rand/1 mutation strategy usually employed a larger crossover rate Cr = 0.9 for optimization problems and it performed very well on multimodal and nonseparable objectives. Furthermore, there were a convention given in the literature “DE/x/y/z” to denote diﬀerent trial vector generation strategies. x denotes the base vector of the donor vector generation strategy, y denotes the number of diﬀerence vector pair, and z denotes the crossover scheme, there are usually two diﬀerent crossover schemes including exponential crossover and the binomial crossover. The binomial crossover usually performed better on real-parameter optimization problems. As we know that the exponential crossover implements a combined 1-point crossover and 2-point crossover, nevertheless, it had an positional bias [3] (also called representative bias) within it because the parameters are unequally selected, then the binomial crossover was proposed to tackle this weakness by treating all parameter equal. However, a selection bias still existed because the potential candidate still unequally selected when employing a ﬁxed crossover rate. That why a QUATRE structure [10–14] was proposed to tackle this selection bias. Generally, these DE variants with ﬁxed control parameters or the canonical QUATRE structure still lacks a good adaptation to the landscape of the objective functions during the evolution, therefore, DE variants with adaptive control parameters [4,5,17–20] or adaptive QUATRE structure [21] became much more welcomed by researchers. Brest et al. [17] proposed a parameter adaptive DE algorithm, called jDE, the control parameters F and Cr were adaptively changed during the evolution. Zhang and Sanderson [18] proposed an adaptive DE with optional external archive, and it won the scale invariant competition at 2008 competition. Tanabe et al. [19] proposed a linear population size reduction based success historical parameter adaptation DE algorithm, and it secured the ﬁrst rank at CEC2014 competition. Meng et al. [5] proposed a parameters with adaptive learning mechanism DE algorithm which tackled some weakness existing in the above mentioned DE variants and secured a better performance. However, the update scheme of crossover rate is heavily dependent on the number of individuals in each group which may fall into a bad adaptation when population size becomes relative small.

An Improved LPALMDE Algorithm

479

In this paper, we proposed an improved Linear population size reduction based Parameters with Adaptive Learning Mechanism Diﬀerential Evolution (iLPALMDE) to enhance the former LPALMDE algorithm. The rest of the paper is organized as follows: Sect. 2 presents a detailed description of the iLPALMDE algorithm. Section 3 presents the experiment analysis under CEC2013 test suit for real-parameter single objective optimization. Finally, the conclusion is given in Sect. 4.

2

The Improved Linear Population Size Reduction Based Parameters with Adaptive Learning Mechanism Diﬀerential Evolution

In this part, we mainly discuss the new improved linear population size reduction based parameters with adaptive learning mechanism diﬀerential evolution. The whole description can be divided into three parts, the ﬁrst part presents the mutation strategy. The second part presents the adaptation schemes for F and Cr. The third part presents the linear population size reduction scheme. 2.1

The Mutation Strategy in iLPALMDE

The Parameters with Adaptive Learning Mechanism employed a mutation strategy with time stamp scheme. The detailed mutation strategy can be found in Eq. 1: p r ,G ) − Xi,G ) + F · (Xr1 ,G − X (1) Vi,G = Xi,G + F · (Xbest,G 2 where Xi,G denotes the target vector of ith individual in the Gth generation, p denotes the top 100p% superior individuals, Xr1 ,G denotes a randomly Xbest,G selected individual for the current population under random selection with r ,G denotes a randomly selected vector from the restriction scheme [4], and X 2 denotes the time union P ∪ A where P denotes the current population and A stamp based external archive. The index r2 is also generated by randomly selection with restriction. F denotes the scale factor and Vi,G denotes the donor vector of the ith individual of the Gth generation. This mutation strategy is directed inherited into the iLPALMDE algorithm. 2.2

The Adaptation Schemes for Control Parameter F and Cr in iLPALMDE

The scale factor F in iLPALMDE algorithm obeys Cauchy distribution, F ∼ C(μF , σF ), and the crossover rate Cr obeys Normal distribution, Cr ∼ N (μCr , σCr ). The initial values of μF and μCr are μF = 0.8, μCr = 0.6 respectively. Both the σF and σCr are set constant values during the evolution, σF = 0.2, σCr = 0.1. All the individuals were divided into k groups by employing stochastic universal selection, and each group are initialized with the

480

Z. Meng et al.

same μF and μCr values. Each individual in the population is associated with its own F and Cr values generated by Eq. 2 and Eq. 3 respectively. Fji = randcj (Fj , σF )

(2)

Crj = randnj (μCr , σCr )

(3)

After initialization, trial vectors can be generated by the mutation strategy with its associated control parameters. If a individual fail to generate a better trial vector, then it was labeled as ‘f ’ individual. Otherwise, it was labeled a ‘s’ individual. The number of ‘s’ and ‘f ’ individuals in the j th group are recorded by nsj and nfj respectively. nsj and nfj satisﬁes a following equation in Eq. 4. ⎧ P (j) · ps = nsj + nfj , ⎪ ⎪ ⎪ ⎪ k ⎪ ⎪ ⎪nSucc = ⎪ (nsj ), ⎪ ⎨ j=1 (4) k ⎪ ⎪ ⎪ ⎪ nF ail = (nfj ), ⎪ ⎪ ⎪ ⎪ j=1 ⎪ ⎩ ps = nSucc + nF ail. where “nSucc” and “nF ail” denote the total number of ‘s’ individuals and ‘f ’ individuals of the population. P (j) denotes the selection probability that a certain individuals was classiﬁed into, and the renewing scheme of P (j) is according to the rule in Eq. 5. ⎧ ⎧ ⎪ ns2j ⎪ ⎨ ⎪ ⎪ ⎪ ⎪rj = nSucc · (nsj + nfj ) , if nsj > 0, ⎨ ⎪ ⎩, otherwise. (5) ⎪ ⎪ r j ⎪ ⎪ ⎪ ⎩P (j) = k (r ) . j=1 j where is a small value, i.e. = 0.01, avoiding null selection probability. After the calculation of selection probability P (j), the crossover rate μCrj and the scale factor μF can be updated according to Eqs. 6 and 7: ⎧ Δfs ⎪ ⎪ ⎪ws = |SCr | ⎪ ⎪ s=1 Δfs ⎪ ⎪ ⎪ ⎪ Δf = f (X i,G ) − f (Ui,G ) ⎪ ⎪ i |SCr | ⎪ 2 ⎪ ⎪ ⎪ s=1 ws · SCr (s) ⎪ mean (S ) = W L Cr ⎪ |SCr | ⎨ s=1 ws · SCr (s) ⎧ (6) ⎪ ⎪ ⎨meanW L (SCr ), if SCr = Ø& max{SCr } > 0 ⎪ ⎪ ⎪ ⎪ μCridx ,G+1 = 0, if SCr = Ø&μCridx ,G = 0 ⎪ ⎪ ⎪ ⎪ ⎩ ⎪ ⎪ . otherwise μ Cridx ,G ⎪ ⎪ ⎪ ⎪ ⎪ randn (μ if μCridx > 0; i Cridx , 0.1), ⎪Cr = ⎪ ⎩ i 0. otherwise

An Improved LPALMDE Algorithm

481

where SCr denotes the set of control parameter Cr values of the ‘s’ individuals and meanW L (SCr ) denotes the weighted Lehmer mean of set SCr . idx denotes the index of a group that has the smallest value of selection probability. ⎧ Δfs ⎪ ⎪ ⎪ws = |SF | ⎪ ⎪ ⎪ s=1 Δfs ⎪ ⎪ ⎪ Δf = f (X i,G ) − f (Ui,G ) ⎪ ⎨ i |SF | 2 (7) s=1 ws · SF (s) mean (S ) = ⎪ W L F |SF | ⎪ ⎪ w · S (s) ⎪ s F s=1 ⎪ ⎪ ⎪ ⎪ meanW L (SF ), if SF = Ø ⎪ ⎪ ⎩μF,G+1 = μF,G , otherwise where SF denotes the set of control parameter F values of the ‘s’ individuals and meanW L denotes the weighted Lehmer mean. 2.3

Linear Population Size Reduction Scheme

The linear population size reduction scheme proposed in LSHADE algorithm [19] was also employed in the iLPLAMDE algorithm, the detailed rule of this reduction scheme is presented in Eq. 8: psG+1 = round[

psmin − psini · nf e + psini ] M axnf e

(8)

where psmin denotes the terminal population size, psini denotes the initial population size, nf emax denotes the maximum number of function evaluations, and nf e denotes the current number of function evaluations. Furthermore, the factor of external archive rhar is set constant during the whole evolution while the size of the external archive changes adaptively according to the dynamic population size, arcsize = rhar × ps. In order to dynamically change the population size of external archive, some worse-ranking individuals are removed from the archive.

3

Experiment Analysis

In this part, we mainly analyze the performance of the novel improved Parameter with Adaptive Learning Mechanism Diﬀerential Evolution (iLPALMDE) algorithm. This new algorithm is veriﬁed under CEC2013 test suite containing 28 benchmark functions. 51 runs are conducted on each function with a ﬁxed maximum number of function evaluation nf emax = 10000·D. The mean and standard deviation are collected from the total 51 runs. Other contrasted DE variants employed their default parameter settings, and the comparison results are given in Table 1. Symbols >, =, < in the parenthesis denotes “Better Performance”, “Similar Performance” and “Worse Performance” respectively. Wilcoxon’s signed rank test with signiﬁcant level α = 0.05 is employed for experiment result evaluation. All these experiments are conducted on a PC with Intel(R) Core(TM)

Table 1. Mean/std ﬁtness error Δf = f − f ∗ comparison on 10D optimization among JADE, SHADE, LSHADE, iLSHADE, jSO, LPALMDE and PaDE is presented here. The results are calculated under 51 independent runs with the ﬁxed maximum number of function evaluations nf emax equalling to 10000D. The overall performance of each algorithm is measured under Wilcoxon’s signed rank test with the signiﬁcant level α = 0.05 in comparison with the new proposed PaDE.

482 Z. Meng et al.

An Improved LPALMDE Algorithm

483

i5-3470 CPU @ 3.2 Hz on RedHat Linux Enterprise Edition 5.5 Operating System and Matlab software with 2011b Unix version. The values of ﬁtness error that smaller than “eps” (eps = 2.2204e−016) are considered as zeros herein. From the table, we can see that the novel iLPALMDE algorithm obtains better or similar performances on 26 benchmarks in comparison with JADE algorithm; it also obtains better or similar performances on 25 benchmarks in comparison with SHADE algorithm; obtains 19 similar or better performances on 20 benchmarks out of 28 benchmarks; obtains 18 better or similar performances out of 28 benchmarks; obtains 20 better or similar performances out of 28 benchmarks. To summarize, the novel iLPALMDE algorithm is competitive with the other contrasted algorithms under CEC2013 test suite.

4

Conclusion

The paper proposed a novel improved linear population size reduction based parameters with adaptive learning mechanism diﬀerential evolution. This novel algorithm proposed a new adaptation scheme for control parameter Cr which tackles the weakness of the former LPALMDE algorithm. This novel iLPALMDE algorithm is veriﬁed under CEC2013 test suites contain 28 benchmarks, and the experiment results show that the iLPALMDE algorithm is competitive with the contrasted DE variants.

References 1. Kennedy, J., Eberhart, R.: Particle swarm optimization. Proc. IEEE Int. Conf. Neural Netw. 4(2), 1942–1948 (1995) 2. Storn, R., Price, K.: Diﬀerential evolution - a simple and eﬃcient adaptive scheme for global optimization over continuous spaces. International Computer Science Institute, Berkeley (1995) 3. Price, K., Storn, R.M., Lampinen, J.A.: Diﬀerential evolution: a practical approach to global optimization. Springer (2006) 4. Pan, J.S., Meng, Z., Xu, H., et al.: A matrix-based implementation of DE algorithm: the compensation and deﬁciency. In: International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, pp. 72–81, Springer, Cham (2017) 5. Meng, Z., Pan, J.-S., Kong, L.: Parameters with adaptive learning mechanism (PALM) for the enhancement of diﬀerential evolution. Knowl. Based Syst. 141, 92–112 (2018) 6. Meng, Z., Pan, J.-S.: A simple and accurate global optimizer for continuous spaces optimization. In: Genetic and Evolutionary Computing, pp. 121–129. Springer (2015) 7. Meng, Z., Pan, J.S., Alelaiwi, A.: A new meta-heuristic ebb-tide-ﬁsh-inspired algorithm for traﬃc navigation. Telecommun. Syst. 62(2), 403–415 (2016) 8. Meng, Z., Pan, J.-S.: Monkey King Evolution: a new memetic evolutionary algorithm and its application in vehicle fuel consumption optimization. Knowl. Based Syst. 97, 144–157 (2016)

484

Z. Meng et al.

9. Pan, J.S., Meng, Z., Chu, S.C., et al.: Monkey King Evolution: an enhanced ebbtide-ﬁsh algorithm for global optimization and its application in vehicle navigation under wireless sensor network environment. Telecommun. Syst. 65(3), 351–364 (2017) 10. Meng, Z., Pan, J.-S., Xu, H.: QUasi-Aﬃne TRansformation Evolutionary (QUATRE) algorithm: a cooperative swarm based algorithm for global optimization. Knowl. Based Syst. 109, 104–121 (2016) 11. Pan, J.S., Meng, Z., Xu, H., et al.: QUasi-Aﬃne TRansformation Evolution (QUATRE) algorithm: a new simple and accurate structure for global, pp. 657–667. Springer (2016) 12. Meng, Z., Pan, J.S., Quasi-aﬃne transformation evolutionary (QUATRE) algorithm: a parameter-reduced diﬀerential evolution algorithm for optimization problems. In: IEEE Congress on Evolutionary Computation (CEC), pp. 4082–4089 (2016) 13. Meng, Z., Pan, J.S.: A Competitive QUasi-Aﬃne TRansformation Evolutionary (C-QUATRE) algorithm for global optimization. In: IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 1644–1649. IEEE (2016) 14. Meng, Z., Pan, J.S.: QUasi-Aﬃne TRansformation Evolutionary (QUATRE) algorithm: the framework analysis for global optimization and application in hand gesture segmentation. In: IEEE 13th International Conference on Signal Processing (ICSP), pp. 1832–1837 (2016) 15. Holland, J.H.: Adaptation in Natural and Artiﬁcial Systems: An Introductory Analysis with Applications to Biology, Control, and Artiﬁcial Intelligence. University of Michigan Press, Ann Arbor (1975) 16. Kirkpatrick, S., Daniel Gelatt, C., Vecchi, M.P.: Optimization by simmulated annealing. Science 220(4598), 671–680 (1983) 17. Brest, J., Greiner, S., Boˇskovi´c, B., Mernik, M., Zumer, V.: Self-adapting control parameters in diﬀerential evolution: a comparative study on numerical benchmark problems. IEEE Trans. Evol. Comput. 10(6), 646–657 (2006) 18. Zhang, J., Sanderson, A.C.: JADE: adaptive diﬀerential evolution with optional external archive. IEEE Trans. Evol. Comput. 13(5), 945–958 (2009) 19. Tanabe, R., Fukunaga, A.S.: Improving the search performance of shade using linear population size reduction. In: IEEE Congress on Evolutionary Computation (CEC), July 2014, pp. 1658–1665 (2014) 20. Meng, Z., Pan, J.-S., Zheng, W.-M.: Diﬀerential evolution utilizing a handful top superior individuals with bionic bi-population structure for the enhancement of optimization performance. Enterp. Inf. Syst. (2018). https://doi.org/10.1080/ 17517575.2018.1491064 21. Meng, Z., Pan, J.-S.: QUasi-Aﬃne TRansformation Evolution with External ARchive (QUATRE-EAR): an enhanced structure for diﬀerential evolution. Knowl. Based Syst. 155, 35–53 (2018)

Author Index

A Ai, Yi, 201 C Cao, Zhongyuan, 394 Chen, Dingjun, 140 Chen, Dong, 268 Chen, J., 448 Chen, Tao, 216, 234 Chen, Wei, 185, 201 Chen, Xiaoyong, 440 Chen, Yaqing, 457 Cui, Yanqu, 193 D Deng, Boer, 216 Deng, Jingchun, 250 Deng, Zhaobin, 394 Du, Guobao, 312, 402 F Feng, Shigang, 394 G Geng, Jingchun, 402 Guo, Xiuyun, 164, 171 H He, Chuan-Lei, 37, 48 He, X., 448 Hongxia, Lv, 327 Hongxia, lv, 369 Hou, Yujie, 457 Hu, Dengfeng, 457 Huang, Gaoyong, 280

Huang, Hao, 59, 353 Huang, Huilin, 164 Huang, Jing, 440 Huang, Shan, 259 Huang, Shuming, 298 Huiying, Jing, 177 Hung, Mao-Hsiung, 465 J Ji, Yufeng, 345 Jiang, Jieying, 67, 76 Jingchun, Geng, 177, 471 Ju, Yanni, 201 K Kuang, Rong, 298, 319 L Lao, Ya-long, 85 Li, Junjie, 171, 208 Li, Junnan, 440 Li, Kang, 171 Li, Shengdong, 140 Li, Wentao, 268, 280 Li, Xiaoqing, 477 Li, Xueting, 193, 402 Li, Yunlong, 430 Li, Zongping, 201 Liang, Fenqiang, 440 Liao, Changyu, 259 Lin, Xinyi, 242 Liu, Bin, 250 Liu, Hongbo, 378, 394 Liu, Lan, 11, 22, 59, 85, 96, 353 Liu, Ling, 225

© Springer Nature Switzerland AG 2019 S. Ni et al. (Eds.): VTCA 2018, SIST 129, pp. 485–487, 2019. https://doi.org/10.1007/978-3-030-04582-1

486 Liu, Lisang, 440 Liu, Su, 76 Liu, Xiaowei, 250 Liu, Yang, 345 Liu, Yingjie, 378 Lu, Mingyu, 394 Lu, Wei-Ke, 11 Luo, Kan, 440 Lv, Hong-xia, 234 Lv, Hongxia, 250 Lv, Hong-Xia, 360 Lv, Miaomiao, 305 Lv, Xiaoyuan, 164 M Ma, Si, 67, 76, 106 Mao, Jian-Nan, 96, 353 Meng, Zhenyu, 477 Miaomiao, Lyu, 177 Minghui, Wang, 335 Mou, Rui-Fang, 133 Mu, Ruifang, 411 N Nguyen, Hoang-Son, 133 Ni, Shaoquan, 124, 140, 185, 193, 268, 280, 298, 305, 312 P Pan, Hongye, 369 Pan, Jeng-Shyang, 477 Pan, Jin-shan, 234 Pan, Jinshan, 242 Pan, Pei-fen, 115 Peng, Xiaoqian, 402 Peng, Zhang, 327, 335 Q Qing, Gong, 148 Qu, Siyuan, 3 S Shaoquan, Ni, 177, 327, 335 Sheen, Maw-Tyan, 420 Shen, Ruyi, 106 Sheng, Tianyi, 3, 156 Shi, Fugen, 216 Song, F., 448 Sun, Zong-Sheng, 48 T Tian, Bowen, 164 Tian, Zhiqiang, 3, 156

Author Index V Vu, Trong-Thuat, 133 Vuong, Xuan-Can, 133 W Wang, Bing, 29, 171, 360 Wang, Chunhui, 216 Wang, Chunzheng, 457 Wang, Lieni, 411 Wang, Lin, 67 Wang, Minghui, 124 Wang, Ping, 225 Wang, Pu, 115 Wang, Qian-ting, 420 Wang, Wei, 394 Wang, Wenxian, 319 Wang, Xiaohong, 345 Wang, Yi-Han, 353 Wang, Yihan, 59 Wang, Yingjie, 115 Wang, Yuanyuan, 386 Wei, Feifei, 208, 369 Wei, Wei, 225 Wen, Di, 208 Weng, Wencai, 394 Wu, Tingting, 298, 319 Wu, Ying, 164 X Xia, Zhou, 327, 335 Xie, Liang, 250 Xie, Xue-Jiao, 96 Xu, Changan, 140, 193 Xue, Feng, 37, 48 Y Yang, Feiyu, 305 Yang, Jinqing, 394 Yang, Kai-yu, 85 Yang, Liping, 378 Yang, Rui, 411 Yang, Yuhua, 124 Yang, Zhenbo, 156 Yao, Shunyu, 22 Yi, Yanni, 242 Yu, Xiao, 48 Yu, Xiuzhen, 411 Yu, Yi-Fan, 96 Z Zhang, Zhang, Zhang, Zhang, Zhang,

Bo, 225, 378 F., 448 Guangyuan, 124 Hai, 193 Hui, 29, 259

Author Index Zhang, Zhang, Zhang, Zhang, Zhang, Zhang,

Jie, 259, 430 Jie-ru, 234 Pingjun, 345 Qiangfeng, 268, 280 Rui, 156 Xuepeng, 312

487 Zhang, Yan, 29 Zhao, Feng, 360 Zhao, Hanyue, 430 Zheng, Junfeng, 430 Zheng, Wei-min, 477 Zuo, Ke, 11

Advances in Smart Vehicular Technology, Transportation, Communication and Applications

Recommend Stories

Idea Transcript

Helpful Links

Smile Life

Get in touch