铁道运输与经济2025，Vol.47Issue(11)：36-51,16.DOI:10.16668/j.cnki.issn.1003-1421.2025.11.03
基于强化学习的铁路列车运行图编制与优化研究综述
Review on Reinforcement Learning-Based Railway Train Working Diagram and Optimization
摘要
Abstract
Train working diagram compilation serves as a core decision-making problem in railway transport organization,demonstrating intrinsic compatibility with Markov decision process modeling in reinforcement learning.The study first theoretically validated the feasibility,advantages,and limitations of applying reinforcement learning to train working diagram compilation tasks and identified key challenges in component design and algorithm development.Through a comprehensive literature review,it was found that current research remained in the exploratory phase.At the core component level,five categories of state space design were summarized,along with action space design approaches based on element modification,process control,and train working diagram compilation actions.Neural network architectures and reinforcement learning algorithms were recommended according to problem characteristics.Technically,the study analyzed the current challenges such as high-dimensional state space representation and multi-dimensional combinatorial space exploration,as well as the breakthrough paths;finally,from an industry perspective,the future technological trends and development directions were prospected.This can provide systematic references for the modeling methods,technical solutions,and industry implementation of subsequent research.关键词
列车运行图编制/强化学习/马尔可夫决策过程/状态空间/动作空间Key words
Train Working Diagram/Reinforcement Learning/Markov Decision Process/State Space/Action Space分类
交通运输引用本文复制引用
陈昂扬,范家铭,徐辉章,齐昕,李博,张新..基于强化学习的铁路列车运行图编制与优化研究综述[J].铁道运输与经济,2025,47(11):36-51,16.基金项目
中国国家铁路集团有限公司科技研究开发计划课题(P2024X002) （P2024X002）
中国铁道科学研究院集团有限公司科研项目(2024YJ154) （2024YJ154）
