基于深度Q网络的高速铁路车站到发线运用冲突调整研究  

Research on Auto Conflict-resolution of Arrival-departure Track Utilization in Large High-speed Railway Stations Based on Deep Q-network

作　　者：田锐 孟令云[1] 王维 李忠灿 唐晓龙 孙飞 TIAN Rui;MENG Lingyun;WANG Wei;LI Zhongcan;TANG Xiaolong;SUN Fei(School of Traffic and Transportation,Beijing Jiaotong University,Beijing 100044,China;Traffic Control Center,China State Railway Group Co.,Ltd.,Beijing 100844,China;Guangzhou Beiyang Information Technology Co.,Ltd.,Guangzhou 510000,China;Department of Automation,Tsinghua University,Beijing 100844,China;China Railway Zhengzhou Group Co.,Ltd.,Zhengzhou 450015,China)

机构地区：[1]北京交通大学交通运输学院,北京100044 [2]中国国家铁路集团有限公司运输调度指挥中心,北京100844 [3]广州北羊信息技术有限公司,广东广州510000 [4]清华大学自动化系,北京100084 [5]中国铁路郑州局集团有限公司,河南郑州450015

出　　处：《铁道学报》2025年第3期1-9,共9页Journal of the China Railway Society

基　　金：中国国家铁路集团有限公司科技研究开发计划(N2022X021);中国铁路郑州局集团有限公司科技研究开发计划(RD2024X006);国家自然科学基金(U2368211,72022003,72401158);中国博士后科学基金(2023M741956)。

摘　　要：高速列车受到自然灾害或设备故障因素干扰时可能导致运行秩序大规模紊乱,引发高速铁路枢纽到发线运用产生冲突、原有方案无法准时执行,需要调整冲突列车的停靠到发线或时刻。目前到发线调整方法主要依靠列车调度员经验,方案调整的效率和质量因人而异。通过对高速铁路枢纽到发线运用冲突类别、消解策略和目标导向进行梳理,构建到发线运用冲突调整优化模型。为解决优化模型求解效率问题,提出深度强化学习框架进行求解。以长沙南站为例,验证深度强化学习框架的求解可行性,结果显示:该模型可以协助快速找到符合运输实际的调整方案,为高速铁路枢纽到发线运用调整提供决策支撑。Operational disruptions in high-speed train systems caused by natural disasters or equipment failures can trigger extensive chaos,leading to conflicts in the utilization of arrival-departure tracks at terminal stations,and failure to implement the original plan on time.Adjustments are required for the arrival-departure tracks or schedule of conflicting trains.At present,the method of adjusting arrival-departure tracks mainly relies on the work experience of train dispatchers,with the efficiency and quality of scheme adjustment varying from person to person.In this paper,the conflict adjustment optimization model for the use of arrival-departure tracks was analyzed by sorting out the conflict categories,resolution strategies and goal orientation in the utilization of arrival-departure tracks at high-speed railway hubs.In order to improve the efficiency of optimization model,a deep-reinforcement learning method was proposed.Based on the case study of Changsha South Railway Station,the feasibility of the method was verified.The results show that the method can help to quickly find an adjustment scheme that meets the actual transportation situation,providing decision-making support for the adjustment of the utilization of arrival-departure tracks at high-speed railway terminal stations.

关 键 词：高速铁路 车站 深度强化学习 决策支撑 

分 类 号：U292.1[交通运输工程—交通运输规划与管理]