基于深度强化学习的高速列车运行节能在线优化

黄子凌; 石红国

doi:10.13238/j.issn.1004-2954.202407290004

English Version

您当前的位置：

首页 >

文章列表页 >

基于深度强化学习的高速列车运行节能在线优化

规划/运输 | 更新时间：2026-06-22

- 基于深度强化学习的高速列车运行节能在线优化
- Energy-Saving Online Optimization of High-Speed Train Operation Based on Deep Reinforcement Learning
- 铁道标准设计 2026年70卷第5期页码：10-19
- 作者机构：
  
  西南交通大学交通运输与物流学院，成都 611756
- 作者简介：
  
  作者简介：黄子凌（2000—），男，硕士研究生，主要从事城轨列车与高速列车运行节能优化研究工作，E-mail: 978291632@qq.com。
  石红国（1974—），男，教授，工学博士，主要从事列车节能运行优化研究工作，E-mail: shg@home.swjtu.edu.cn。
- 基金信息：
  
  国家重点研发计划项目(2022YFB4300504-1);四川省自然科学基金项目(2022NSFSC0418;2023NSFSC0386)
- DOI：10.13238/j.issn.1004-2954.202407290004
  中图分类号： U238;TP181
- 收稿：2024-07-29，
  
  修回：2024-09-06，
  
  网络首发：2025-03-11，
  
  纸质出版：2026-05-20
- 稿件说明：
移动端阅览
黄子凌, 石红国. 基于深度强化学习的高速列车运行节能在线优化[J]. 铁道标准设计, 2026,70(5):10-19.

HUANG Ziling, SHI Hongguo. Energy-Saving Online Optimization of High-Speed Train Operation Based on Deep Reinforcement Learning[J]. Railway Standard Design, 2026, 70(5): 10-19.
黄子凌, 石红国. 基于深度强化学习的高速列车运行节能在线优化[J]. 铁道标准设计, 2026,70(5):10-19. DOI： 10.13238/j.issn.1004-2954.202407290004.

HUANG Ziling, SHI Hongguo. Energy-Saving Online Optimization of High-Speed Train Operation Based on Deep Reinforcement Learning[J]. Railway Standard Design, 2026, 70(5): 10-19. DOI： 10.13238/j.issn.1004-2954.202407290004.

摘要

【目的】

高速列车运行节能优化研究对于铁路经济效益和节能减排均有重大意义。

【方法】

首先，采用划分子区间的方法对线路进行预处理，然后在考虑再生制动、分相区、长陡坡、变限速等多个因素的基础上构建列车节能操纵模型，并将模型转换为基于子区间分步优化的马尔可夫决策过程。为求解马尔可夫决策过程，引入深度强化学习算法，提出一种基于SAC（Soft Actor-Critic）算法的在线节能优化方法。

【结果】

选取实际线路对算法进行有效性验证，结果表明，在满足约束条件的前提下，SAC算法较PPO（Proximal Policy Optimization）算法具有更快的收敛效率，较粒子群算法与PPO算法的节能效果更佳，节能率达到37.4%。同时，选取线路中若干个位置进行仿真试验，结果表明算法均能实时生成剩余路线的节能运行曲线，验证了算法的在线优化能力。【结论】研究成果为高速列车运行节能优化提供依据。

Abstract

[

Objective

]

The research on energy-saving optimization of high-speed trains is of great significance for the economic benefits of railway operations and energy conservation and emission reduction. [

Methods

]

First

the railway line was preprocessed by subdividing it into several subintervals. Then

considering factors such as regenerative braking

neutral sections

long and steep slopes

and variable speed limits

a train energy-saving control model was constructed. This model was then transformed into a Markov decision process based on stepwise optimization of the subintervals. To solve the Markov decision process

a deep reinforcement learning algorithm was introduced

and an online energy optimization method based on the soft actor-critic（SAC）algorithm was proposed. [

Results

]

The algorithm's effectiveness was validated using an actual railway line

and the results showed that

under the given constraints

the SAC algorithm converged faster than the proximal policy optimization（PPO）algorithm and achieved better energy savings compared to both the particle swarm optimization（PSO）and PPO algorithms

with an energy-saving rate of 37.4%. Additionally，simulations conducted at several locations along the route indicated that the algorithm generated energy-efficient speed profiles for the remaining route in real time

confirming its online optimization capability. [

Conclusion

]

The research findings provide a basis for energy-saving optimizatio a high speed train operations.

关键词

Keywords

references

ICHIKAWA K. Application of Optimization Theory for Bounded State Variable Problems to the Operation of Train[J]. Bulletin of JSME, 1968,11（47）:857-865.

PHIL H. An Optimal Strategy for the Control of a Train[J]. The Journal of the Australian Mathematical Society Series B Applied Mathematics, 1990,31（4）:454-471.

PHIL H. Optimal Strategies for the Control of a Train[J]. Automatica, 1996,32（4）:519-532.

CHENG J X, HOWLETT P. A Note on the Calculation of Optimal Strategies for the Minimization of Fuel Consumption in the Control of Trains[J]. IEEE Transactions on Automatic Control, 1993,38（11）:1730-1734.

HOWLETT P G, CHENG J. Optimal Driving Strategies for a Train on a Track with Continuously Varying Gradient[J]. The Journal of the Australian Mathematical Society Series B Applied Mathematics, 1997,38（3）:388-410.

PHIL H. The Optimal Control of a Train[J]. Annals of Operations Research, 2000,98（1）:65-87.

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

TRIVELLA A, WANG P L, CORMAN F. The Impact of Wind on Energy-efficient Train Control[J]. EURO Journal on Transportation and Logistics, 2021,10:100013.

WANG X K, TANG T, SU S, et al. An Integrated Energy-efficient Train Operation Approach Based on the Space-time-speed Network Methodology[J]. Transportation Research Part E: Logistics and Transportation Review, 2021, 150: 102323.

XING Z Y, ZHANG Z Y, GUO J, et al. Rail Train Operation Energy-saving Optimization Based on Improved Brute-force Search [J]. Applied Energy, 2023, 330: 120345.

LI W, ZHAO S Z, LI K, et al. GSOANR-based Multi-objective Train Trajectory Optimization[J]. International Journal of Rail Transportation, 2024,12（4）:733-748.

HE D Q, LU G C, YANG Y J. Research on Optimization of Train Energy-saving Based on Improved Chicken Swarm Optimization[J]. IEEE Access, 2019,7:121675-121684.

LU G C, HE D Q, ZHANG J L. Energy-saving Optimization Method of Urban Rail Transit Based on Improved Differential Evolution Algorithm[J]. Sensors, 2022,23（1）:378.

LIU K, WANG X, WANG L. An Improved Memetic Algorithm for Urban Rail Train Operation Strategy Optimization [J]. International Journal of Innovative Computing Information and Control, 2020, 16（1）:241-256.

WANG Y H, DE SCHUTTER B, VAN DEN BOOM T J J, et al. Optimal Trajectory Planning for Trains-a Pseudospectral Method and a Mixed Integer Linear Programming Approach[J]. Transportation Research Part C: Emerging Technologies, 2013,29:97-114.

FENG M L, WU C X, LU S F, et al. Notch-based Speed Trajectory Optimisation for High-speed Railway Automatic Train Operation[J]. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 2022,236（2）:159-171.

WEI G D, ZHU S W, WANG Y H, et al. Energy-efficient Automatic Train Operation for High-speed Railways: Considering Discrete Notches and Neutral Sections[J]. Transportation Research Part C: Emerging Technologies, 2022,145:103884.

GU Q, TANG T, CAO F, et al. Energy-efficient Train Operation in Urban Rail Transit Using Real-time Traffic Information [J]. IEEE Transactions on Intelligent Transportation Systems, 2014, 15（3）:1216-1233.

SU S, ZHU Q Y, LIU J Q, et al. A Data-driven Iterative Learning Approach for Optimizing the Train Control Strategy[J]. IEEE Transactions on Industrial Informatics, 2023,19（7）:7885-7893.

孟建军，蒋小一，陈晓强，等．基于Sarsa算法的城轨列车节能控制策略研究[J]．铁道标准设计，2024,68（8）:8-14.

MENG Jianjun, JIANG Xiaoyi, CHEN Xiaoqiang, et al. Intelligent Control Strategy of Urban Rail Train Based on Sarsa Algorithm[J]. Railway Standard Design, 2024,68（8）:8-14.

周敏，董海荣，周学影，等．临时限速下基于强化学习的高速列车速度曲线优化[J]．铁道学报，2023,45（2）:84-92.

ZHOU Min, DONG Hairong, ZHOU Xueying, et al. Reinforcement Learning-based Optimization of Speed Profile for High-speed Train with Temporary Speed Restriction[J]. Journal of the China Railway Society, 2023,45（2）:84-92.

LIU X H, ZHANG Y D, GUO J. Research on Optimization of Energy-saving Driving Strategy of High-speed Train Based on PPO[C]∥2023 3rd International Conference on Computer, Control and Robotics（ICCCR）. Shanghai, China. IEEE, 2023:258-264.

ZHANG Z Y, CHENG X Q, XING Z Y, et al. Pareto Multi-objective Optimization of Metro Train Energy-saving Operation Using Improved NSGA-Ⅱ Algorithms [J]. Chaos, Solitons & Fractals, 2023,176:114183.

HAARNOJA T, ZHOU A, ABBEEL P, et al. Soft Actor-Critic:Off-policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor [EB/OL]. 2018: arXiv: 1801.01290.https://arxiv. org/abs/1801.01290.

HAARNOJA T, ZHOU A, HARTIKAINEN K, et al. Soft Actor-critic Algorithms and Applications[EB/OL]. 2018: arXiv: 1812.05905.https://arxiv. org/abs/1812.05905.

FERNÁNDEZ-RODRÍGUEZ A, SU S, FERNÁNDEZ-CARDADOR A, et al. A Multi-objective Algorithm for Train Driving Energy Reduction with Multiple Time Targets[J]. Engineering Optimization，2021,53（4）:719-734.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

强风沙环境下列车气动特性模拟研究

侧风环境下城轨列车自动驾驶控制策略研究

高速列车-悬索桥随机振动的行车安全概率评判研究

横风作用下基于PSO寻优的高速列车PID主动悬挂控制器设计

基于特征融合变维卷积神经网络的高铁轮轨不良状态识别方法