基于XGBoost的城市轨道列车ATO运行控制方法研究
Research on Urban Rail Train ATO Operation Control Method Based on XGBoost
DOI: 10.12677/airr.2026.155104, PDF,   
作者: 董亚琨：兰州交通大学光电技术与智能控制教育部重点实验室，甘肃 兰州
关键词: 城市轨道交通；自动列车运行；XGBoost；PID控制；经验规则控制；节能控制；Urban Rail Transit； Automatic Train Operation； XGBoost； Pid Control； Rule-Based Control； Energy-Saving Control
摘要: 针对城市轨道列车ATO运行中准点性、停车精度、舒适性与节能性难以兼顾的问题，提出基于XGBoost的加速度指令预测控制方法。以次渠区段为主要实验场景，建立考虑线路限速、坡度、列车动力学、执行机构动态和再生制动储能的仿真模型，构建剩余距离、时间比例、线路状态、推荐速度及速度滑动窗口等特征，采用XGBoost在线预测加速度指令。测试集MAE、RMSE和R2分别为0.0409 m/s2、0.1463 m/s2和0.9132。次渠区段仿真结果表明，XGBoost控制器实际能耗为41.29 kJ，较经验规则和PID分别降低40.89%和28.42%，平均冲击率为0.0810 m/s3，PID能量回收率最高，为79.28%；经验规则控制器运行时间最短，为115.40 s。宋家庄补充线路实验表明，该方法在复杂坡度和多级限速条件下仍具有一定节能潜力，但停车精度和跨线路适应性仍需提高。结果表明，所提方法可降低实际能耗并改善运行平稳性，可为城市轨道列车数据驱动ATO节能控制提供参考。
Abstract: An XGBoost-based acceleration command prediction method is proposed to coordinate punctuality, stopping accuracy, comfort and energy saving in urban rail ATO. A Ciqu-section simulation model is established considering speed limits, gradients, train dynamics, actuator dynamics and regenerative braking energy storage. Running-state features, including remaining distance, time ratio, line state, recommended speed and sliding-window speed, are used to train an XGBoost regression model. The test-set MAE, RMSE and R2 are 0.0409 m/s2, 0.1463 m/s2 and 0.9132. In the Ciqu simulation, XGBoost consumes 41.29 kJ, 40.89% and 28.42% lower than the rule-based and PID controllers, with the lowest average jerk of 0.0810 m/s3. The PID controller has the highest energy recovery rate of 79.28%, while the rule-based controller has the shortest running time of 115.40 s. Supplementary Songjiazhuang results show that this method still has energy-saving potential under complex gradients and multiple speed limits, while stopping accuracy and cross-line adaptability need improvement. The results show that the proposed method can reduce actual energy consumption and improve running smoothness, and can provide a reference for data-driven ATO energy-saving control of urban rail transit trains.
文章引用：董亚琨. 基于XGBoost的城市轨道列车ATO运行控制方法研究[J]. 人工智能与机器人研究, 2026, 15(5): 1143-1155. https://doi.org/10.12677/airr.2026.155104

参考文献

[1] 余祖俊, 王洪伟, 王悉, 等. 轨道交通自主运行控制技术综述[J]. 北京交通大学学报, 2025, 49(5): 6-33.
[2] 高豪, 张亚东, 郭进, 李科宏. 基于动态规划的列车节能运行两阶段优化方法[J]. 西南交通大学学报, 2020, 55(5): 946-954.
[3] 冉昕晨, 陈绍宽, 柏赟, 等. 应对潮汐客流的城市轨道交通列车节能和乘客节时运行图优化模型[J]. 中国铁道科学, 2022, 43(1): 171-181.
[4] De Martinis, V. and Corman, F. (2018) Data-Driven Perspectives for Energy Efficient Operations in Railway Systems: Current Practices and Future Opportunities. Transportation Research Part C: Emerging Technologies, 95, 679-697.
https://doi.org/10.1016/j.trc.2018.08.008
[5] Wang, P. and Goverde, R.M.P. (2019) Multi-Train Trajectory Optimization for Energy-Efficient Timetabling. European Journal of Operational Research, 272, 621-635.
https://doi.org/10.1016/j.ejor.2018.06.034
[6] Zhang, H., Jia, L., Wang, L. and Xu, X. (2019) Energy Consumption Optimization of Train Operation for Railway Systems: Algorithm Development and Real-World Case Study. Journal of Cleaner Production, 214, 1024-1037.
https://doi.org/10.1016/j.jclepro.2019.01.023
[7] Huang, K., Wu, J., Yang, X., Gao, Z., Liu, F. and Zhu, Y. (2019) Discrete Train Speed Profile Optimization for Urban Rail Transit: A Data-Driven Model and Integrated Algorithms Based on Machine Learning. Journal of Advanced Transportation, 2019, Article ID: 7258986.
https://doi.org/10.1155/2019/7258986
[8] Chen, T. and Guestrin, C. (2016) XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, 13-17 August 2016, 785-794.
https://doi.org/10.1145/2939672.2939785
[9] Wang, X., Li, S., Su, S. and Tang, T. (2019) Robust Fuzzy Predictive Control for Automatic Train Regulation in High-Frequency Metro Lines. IEEE Transactions on Fuzzy Systems, 27, 1295-1308.
https://doi.org/10.1109/tfuzz.2018.2877593
[10] Xun, J., Yin, J., Liu, R., Liu, F., Zhou, Y. and Tang, T. (2019) Cooperative Control of High-Speed Trains for Headway Regulation: A Self-Triggered Model Predictive Control Based Approach. Transportation Research Part C: Emerging Technologies, 102, 106-120.
https://doi.org/10.1016/j.trc.2019.02.023
[11] Su, S., Wang, X., Tang, T., Wang, G. and Cao, Y. (2021) Energy-Efficient Operation by Cooperative Control among Trains: A Multi-Agent Reinforcement Learning Approach. Control Engineering Practice, 116, Article ID: 104901.
https://doi.org/10.1016/j.conengprac.2021.104901
[12] Li, S., Liu, R., Gao, Z. and Yang, L. (2021) Integrated Train Dwell Time Regulation and Train Speed Profile Generation for Automatic Train Operations on High-Density Metro Lines: A Distributed Optimal Control Method. Transportation Research Part B: Methodological, 148, 82-105.
https://doi.org/10.1016/j.trb.2021.04.009
[13] Cheng, Y., Yin, J. and Yang, L. (2021) Robust Energy-Efficient Train Speed Profile Optimization in a Scenario-Based Position-Time-Speed Network. Frontiers of Engineering Management, 8, 595-614.
https://doi.org/10.1007/s42524-021-0173-1
[14] Shang, M., Zhou, Y. and Fujita, H. (2021) Deep Reinforcement Learning with Reference System to Handle Constraints for Energy-Efficient Train Control. Information Sciences, 570, 708-721.
https://doi.org/10.1016/j.ins.2021.04.088
[15] He, D., Zhang, L., Guo, S., Chen, Y., Shan, S. and Jian, H. (2021) Energy-Efficient Train Trajectory Optimization Based on Improved Differential Evolution Algorithm and Multi-Particle Model. Journal of Cleaner Production, 304, Article ID: 127163.
https://doi.org/10.1016/j.jclepro.2021.127163
[16] Li, J., Pan, F., Tang, H., Tong, S., Zhang, L., Li, X., et al. (2022) Energy-Saving Metro Train Timetable Optimization Method Based on a Dynamic Passenger Flow Distribution. Journal of Advanced Transportation, 2022, Article ID: 9776845.
https://doi.org/10.1155/2022/9776845
[17] Zhang, L., He, D., He, Y., Liu, B., Chen, Y. and Shan, S. (2022) Real-Time Energy Saving Optimization Method for Urban Rail Transit Train Timetable under Delay Condition. Energy, 258, Article ID: 124853.
https://doi.org/10.1016/j.energy.2022.124853
[18] Li, G., Or, S.W. and Chan, K.W. (2023) Intelligent Energy-Efficient Train Trajectory Optimization Approach Based on Supervised Reinforcement Learning for Urban Rail Transits. IEEE Access, 11, 31508-31521.
https://doi.org/10.1109/access.2023.3261900
[19] Xing, Z., Zhang, Z., Guo, J., Qin, Y. and Jia, L. (2023) Rail Train Operation Energy-Saving Optimization Based on Improved Brute-Force Search. Applied Energy, 330, Article ID: 120345.
https://doi.org/10.1016/j.apenergy.2022.120345
[20] Chen, Z., Li, S. and Yang, L. (2023) Hierarchical Optimal Control Framework to Automatic Train Regulation Combined with Energy-Efficient Speed Trajectory Calculation in Metro Lines. Transportation Research Part C: Emerging Technologies, 149, Article ID: 104059.
https://doi.org/10.1016/j.trc.2023.104059
[21] Yue, L., Liu, L., Li, M., Xiao, B. and Wu, X. (2023) Research on Text Fault Recognition for On-Board Equipment of a C3 Train Control System Based on an Integrated XGBoost Algorithm. Transportation Safety and Environment, 5, tdac066.
https://doi.org/10.1093/tse/tdac066
[22] Zhou, M., Hou, Z., Wu, X., Dong, H. and Wang, F. (2024) Integration of Train Regulation and Speed Profile Optimization Based on Feature Learning and Hybrid Search Algorithm. IEEE Transactions on Computational Social Systems, 11, 2535-2544.
https://doi.org/10.1109/tcss.2023.3303473
[23] Yuan, Y., Li, S., Yang, L. and Gao, Z. (2024) Nonlinear Model Predictive Control to Automatic Train Regulation of Metro System: An Exact Solution for Embedded Applications. Automatica, 162, Article ID: 111533.
https://doi.org/10.1016/j.automatica.2024.111533
[24] Li, S., Yuan, Y., Chen, Z., Yang, L. and Yu, C. (2024) Real-Time Train Regulation in the Metro System with Energy Storage Devices: An Efficient Decomposition Algorithm with Bound Contraction. Transportation Research Part C: Emerging Technologies, 159, Article ID: 104493.
https://doi.org/10.1016/j.trc.2024.104493
[25] Sun, Z., He, D., He, Y., Shan, S. and Zhou, J. (2024) A Bi-Objective Optimization Model of Metro Trains Considering Energy Conservation and Passenger Waiting Time. Journal of Cleaner Production, 437, Article ID: 140427.
https://doi.org/10.1016/j.jclepro.2023.140427
[26] Zhong, L., Xu, G. and Liu, W. (2024) Energy-Efficient and Demand-Driven Train Timetable Optimization with a Flexible Train Composition Mode. Energy, 305, Article ID: 132183.
https://doi.org/10.1016/j.energy.2024.132183
[27] Wang, Z., Quaglietta, E., Bartholomeus, M.G.P., Cunillera, A. and Goverde, R.M.P. (2025) Optimising Timing Points for Effective Automatic Train Operation. Computers & Industrial Engineering, 206, Article ID: 111237.
https://doi.org/10.1016/j.cie.2025.111237
[28] Lian, D., Chen, Z., Mo, P., Gao, Z., D’Ariano, A. and Yang, L. (2025) Energy-Efficient Multi-Curve Optimization in Urban Rail Transit: Stability Enhancement under Operational Uncertainties and Curve Adjustments. Transportation Research Part C: Emerging Technologies, 176, Article ID: 105148.
https://doi.org/10.1016/j.trc.2025.105148
[29] Wang, D., Wu, J., Chang, X. and Yin, H. (2025) Distributed Multi-Agent Reinforcement Learning Approach for Energy-Saving Optimization under Disturbance Conditions. Transportation Research Part E: Logistics and Transportation Review, 200, Article ID: 104180.
https://doi.org/10.1016/j.tre.2025.104180
[30] Wang, D., Wu, J., Wei, Y., et al. (2024) Energy-Saving Operation in Urban Rail Transit: A Deep Reinforcement Learning Approach with Speed Optimization. Travel Behaviour and Society, 35, Article ID: 100796.
[31] Wang, X., D’Ariano, A., Su, S. and Tang, T. (2023) Cooperative Train Control during the Power Supply Shortage in Metro System: A Multi-Agent Reinforcement Learning Approach. Transportation Research Part B: Methodological, 170, 244-278.
https://doi.org/10.1016/j.trb.2023.02.015

为你推荐