基于交互式仿真和代理模型的高铁列车调度  

Interactive Simulator and Surrogate Model-based High-speed Railway Train Timetable Scheduling

作　　者：刘瑞广 崔东亮[1] 代学武 袁志明 杨鹏鑫 王星皓 LIU Ruiguang;CUI Dongliang;DAI Xuewu;YUAN Zhiming;YANG Pengxin;WANG Xinghao(State Key Laboratory of Synthetical Automation for Process Industries,Northeastern University,Shenyang 110819,China;Signal&Communication Research Institute,China Academy of Railway Sciences Co.,Ltd.,Beijing 100081,China)

机构地区：[1]东北大学流程工业综合自动化全国重点实验室,辽宁沈阳110819 [2]中国铁道科学研究院集团有限公司通信信号研究所,北京100081

出　　处：《控制工程》2025年第3期447-458,共12页Control Engineering of China

基　　金：辽宁省自然科学基金资助项目(2020-MS-093);国家自然科学基金资助项目(61790574,U1834211);兴辽计划(XLYC1808001)。

摘　　要：突发事件下交汇站点的多区段列车协同调度中,由于各区段列车运行过程相互影响,难以用数学模型描述,且重调度方案可行性差。针对该问题,提出了一种交互式中观仿真平台、代理模型和启发式算法相结合的高铁智能调度优化方法。首先,设计了交互式中观仿真平台,可根据实际线路对列车运行进行真实推演,以获取复杂区段重调度方案,并提供了更加真实的评价指标。其次,为进一步克服仿真计算量大导致难以满足快速决策的需求,提出了基于主动学习的高斯过程模型优化方法,构建了列车晚点增量代理模型,可有效降低模型构建的计算代价。最后,以沈阳枢纽站为场景,以最小化多区段列车总晚点和晚点增量为优化目标进行了仿真实验,结果表明,所提方法能获得多区段路网更优的重调度方案。The cooperative scheduling of trains in multiple sections at handover stations under emergency events is difficult to be described by mathematical models because the train operation process of each section affects each other,and the feasibility of rescheduling schemes obtained is poor.In response to the question,an interactive mesoscopic simulation platform,surrogate model and heuristic algorithm combined with intelligent scheduling optimization method is proposed for high-speed railway.The interactive mesoscopic simulation platform is designed,which can be used to truly deduce the operation process of trains according to the actual route layout.The approach can obtain rescheduling schemes of complex sections,and provide a more true and comprehensive evaluation index.In order to further overcome the problem of large amount of simulation calculation,which is difficult to meet the demand for rapid decision making,an optimization method of Gaussian process model based on active learning is proposed.In this way,the surrogate model of high-speed train delay increment is constructed,which can effectively reduce the computational cost of model construction.The proposed method is validated using Shenyang handover station as the experimental scenario with the objective of minimizing the total delay and incremental delay of trains in multiple sections.The experimental results show that the proposed scheduling optimization method can obtain a better rescheduling scheme for a multi-section railway network.

关 键 词：多区段调度 代理模型 交互式仿真 复杂路网 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]