考虑负载不确定性的城轨列车目标速度曲线实时鲁棒优化

doi:10.16097/j.cnki.1009-6744.2025.04.023

交通运输系统工程与信息 ›› 2025, Vol. 25 ›› Issue (4): 254-264.DOI: 10.16097/j.cnki.1009-6744.2025.04.023

考虑负载不确定性的城轨列车目标速度曲线实时鲁棒优化

朱琴跃^*，李纪元，李泓羿，钱舒杨，赵亚辉

同济大学，电子与信息工程学院，上海201804

收稿日期:2025-03-07 修回日期:2025-05-26 接受日期:2025-06-03 出版日期:2025-08-25 发布日期:2025-08-25
作者简介:朱琴跃(1970—)，女，江苏无锡人，教授，博士。

Real-time Robust Optimization of Target Speed Profiles for Urban Rail Trains Considering Load Uncertainty

ZHU Qinyue^*, LI Jiyuan, LI Hongyi, QIAN Shuyang, ZHAO Yahui

School of Electronics and Information Engineering, Tongji University, Shanghai 201804, China

Received:2025-03-07 Revised:2025-05-26 Accepted:2025-06-03 Online:2025-08-25 Published:2025-08-25

摘要/Abstract

摘要： 针对城轨列车自动驾驶过程中乘客负载不确定性变化对列车正常运行的影响，本文提出一种考虑负载不确定性的列车目标速度曲线实时鲁棒优化方法，包括模型设计、模型训练和模型验证这3个部分。首先，基于马尔可夫决策过程构建列车驾驶强化学习模型，其奖励设计兼顾对性能指标和操纵策略的鲁棒优化；其次，使用基于潜力奖励塑形技术(Potential-Based Reward Shaping,PBRS)改善模型训练的收敛性能，通过深度Q网络(DeepQ-Network, DQN)估计价值函数，实现实时响应城轨列车负载的变化；最后，以北京地铁某运营线路的列车运行场景为仿真案例验证模型的有效性。仿真结果表明，DQN-PBRS算法的平均计算时长为26ms，可实现实时生成列车的目标速度，生成的目标速度曲线在极端负载和负载变化情况下相较于DQN算法具有更好的鲁棒性，且列车运行能耗降低5%以上。通过对算法中关键超参数进行敏感性分析，确定了训练效果最优的超参数组合。

关键词: 铁路运输, 目标速度曲线优化, 深度强化学习, 城轨列车, 负载不确定性

Abstract: To address the impact of uncertain passenger load variations on train operation in urban rail transit, this paper proposed a real-time robust optimization method for target speed profiles considering load uncertainty. The implementation included three parts: model design, model training, and model validation. First, a reinforcement learning model for train operation was developed based on the Markov decision process, with reward design balancing the robust optimization of performance metrics and control strategies. Second, the model training convergence performance was enhanced by employing the Potential-Based Reward Shaping (PBRS) technology. Real-time response to passenger load changes was achieved through the Deep Q-Network (DQN) value function estimation. At last, the effectiveness of the model was validated via simulation cases based on train operation scenarios of a Beijing subway line. The simulation results show that the DQN-PBRS algorithm achieves an average computation time of 26 millisecond, enabling real-time generation of target speeds. The generated speed profiles exhibit better robustness under extreme load and load variation conditions compared to the DQN algorithm, while also reducing energy consumption by more than 5%. By conducting a sensitivity analysis of key hyperparameters in the algorithm, the optimal hyperparameter combination for the best training performance was determined.

Key words: railway transportation, target speed profile optimization, deep reinforcement learning, urban rail train, load uncertainty

中图分类号:

U268.4

朱琴跃, 李纪元, 李泓羿, 钱舒杨, 赵亚辉. 考虑负载不确定性的城轨列车目标速度曲线实时鲁棒优化[J]. 交通运输系统工程与信息, 2025, 25(4): 254-264.

ZHU Qinyue, LI Jiyuan, LI Hongyi, QIAN Shuyang, ZHAO Yahui. Real-time Robust Optimization of Target Speed Profiles for Urban Rail Trains Considering Load Uncertainty[J]. Journal of Transportation Systems Engineering and Information Technology, 2025, 25(4): 254-264.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: http://www.tseit.org.cn/CN/10.16097/j.cnki.1009-6744.2025.04.023

http://www.tseit.org.cn/CN/Y2025/V25/I4/254

参考文献

[1] NING L, ZHOU M, HOU Z, et al. Deep deterministic policy gradient for high-speed train trajectory optimization[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(8): 11562-11574.

[2] LU S, HILLMANSEN S, HO T K, et al.Single-train trajectory optimization[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 14(2): 743 750.

[3] 赵东升,赵鹏,姚向明,等.基于工况序列寻优的列车节能操纵策略优化[J]. 交通运输系统工程与信息, 2024, 24(2): 157-165. [ZHAO D S, ZHAO P, YAO X M. An energy-efficient train driving strategy based on regime sequences optimization[J]. Journal of Transportation Systems Engineering and Information Technology, 2024, 24(2): 157-165.]

[4] CAO Y, WANG Z C, LIUF, et al. Bio-inspired speed curve optimization and sliding mode tracking control for subway trains[J]. IEEE Transactions on Vehicular Technology, 2019, 68(7): 6331-6342.

[5]林俊亭,李茂林,邱晓辉.延误场景下列车速度曲线与动态调度联合优化方法[J].交通运输系统工程与信息, 2025, 25(1): 173-187. [LIN J T, LI M L, QIU X H. A joint optimization method of train speed curves and dynamic scheduling under delay scenarios[J]. Journal of Transportation Systems Engineering and Information Technology, 2025, 25(1): 173-187.]

[6]楚彭子,袁建军,陈义军.常导高速磁浮列车节能速度曲线鲁棒优化研究[J].铁道科学与工程学报,2023,20 (11): 4062- 4073. [CHU P Z, YUAN J J, CHEN Y J. Robust optimization of energy-efficient speed profile for normal high-speed maglev[J]. Journal of Railway Science and Engineering, 2023, 20(11): 4062-4073.]

[7] CHEN X, LI K, ZHANG L, et al. Robust optimization of energy-saving train trajectories under passenger load uncertainty based on P-NSGA-II[J]. IEEE Transactions on Transportation Electrification, 2022, 9(1): 1826-1844.

[8] FERNANDEZ-RODRIGUEZ A, FERNANDEZ CARDADOR A, CUCALA A P, et al. Design of robust and energy-efficient ATO speed profiles of metropolitan lines considering train load variations and delays[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(4): 2061-2071.

[9] WANG L, YANG L, GAO Z, et al. Robust train speed trajectory optimization: A stochastic constrained shortest path approach[J]. Frontiers of Engineering Management, 2017, 4(4): 408-417.

[10] ZHOU K, SONG S, XUE A, et al. Smart train operation algorithms based on expert knowledge and reinforcement learning[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2020, 52(2): 716-727.

[11] LI G, OR S W, CHAN K W. Intelligent energy-efficient train trajectory optimization approach based on supervised reinforcement learning for urban rail transits [J]. IEEE Access, 2023, 11: 31508-31521.

[12] HOBEROCK L L. A survey of longitudinal acceleration comfort studies in ground transportation vehicles[R]. Austin: Council for Advanced Transportation Studies, 1976.

[13] TAYLOR M E, STONE P. Transfer learning for reinforcement learning domains: A survey[J]. Journal of Machine Learning Research, 2009, 10(7): 1633-1685.

[14] DEVLIN S, KUDENKO D. Theoretical considerations of potential-based reward shaping for multi-agent systems [C]. Taipei: Tenth International Conference on Autonomous Agents and Multi-Agent Systems, 2011.

[1]	代亮, 杜鹏飞, 黄自彬, 杨朋博. 基于深度强化学习的城市交通信号分层协同控制方法[J]. 交通运输系统工程与信息, 2025, 25(4): 63-72.
[2]	陈喜春, 杨阳, 田小鹏. 城际与市域铁路过轨运输下列车时刻表和停站方案协同优化[J]. 交通运输系统工程与信息, 2025, 25(4): 241-253.
[3]	李嘉炜, 田大新, 吴思凡. 不对称可变时滞下的列车虚拟编组稳定控制方法[J]. 交通运输系统工程与信息, 2025, 25(4): 265-274.
[4]	高鹏飞, 郑伟, 王洪伟, 李晔. 基于多层网络加权投影的站内关键轨道区段识别研究[J]. 交通运输系统工程与信息, 2025, 25(4): 275-286.
[5]	周文梁, 朱翚鸿, 光可欣, 郭一唯. 基于优先级排序的高铁列车时刻表优化[J]. 交通运输系统工程与信息, 2025, 25(4): 287-296.
[6]	杜剑, 秦可萱, 林姗, 张然, 李洋, 杨忠杰. 考虑集结模式的中欧班列开行方案与运行图联合优化[J]. 交通运输系统工程与信息, 2025, 25(3): 61-72.
[7]	田志强, 刘磊, 孙国锋, 张俊峰, 梁辉. 铁路长大坡道线路考虑维修天窗设置的列车运行图优化[J]. 交通运输系统工程与信息, 2025, 25(3): 288-298.
[8]	帅斌, 刘一江, 许旻昊, 谢安昊, 孙宗胜, 法慧妍. 高铁快运专列运行图加线方法研究[J]. 交通运输系统工程与信息, 2025, 25(3): 276-287.
[9]	南斯睿, 于谦, 李铁柱, 尚赞娣, 陈海波. 考虑进站策略的网联电动公交车节能驾驶优化研究[J]. 交通运输系统工程与信息, 2025, 25(2): 82-94.
[10]	代亮, 黄自彬, 张中昊, 李臣富. 考虑车道剩余容量的区域交通信号控制方法[J]. 交通运输系统工程与信息, 2025, 25(2): 108-118.
[11]	苏焕银, 莫尚霖, 代慧子. 城市轨道交通拟动态随机用户均衡分配方法[J]. 交通运输系统工程与信息, 2025, 25(2): 273-281.
[12]	龚帅宇, 徐行方, 鲁玉. 省域铁路成网条件下列车开行方案优化[J]. 交通运输系统工程与信息, 2025, 25(2): 282-292.
[13]	冯芬玲, 方源, 张泽, 董开云. 风险扰动视角下中欧班列运输网络韧性评估[J]. 交通运输系统工程与信息, 2025, 25(2): 338-351.
[14]	林俊亭, 李茂林, 邱晓辉. 延误场景下列车速度曲线与动态调度联合优化方法[J]. 交通运输系统工程与信息, 2025, 25(1): 173-187.
[15]	吴刚, 杨锋, 江山, 郭茜. 基于霍特林模型的中欧班列最佳补贴值[J]. 交通运输系统工程与信息, 2025, 25(1): 289-297.

考虑负载不确定性的城轨列车目标速度曲线实时鲁棒优化

Real-time Robust Optimization of Target Speed Profiles for Urban Rail Trains Considering Load Uncertainty

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics