基于深度强化学习的铁路更新维护计划  

Railway update and maintenance plan based on Deep Reinforcement Learning

作　　者：白宏权 吕和平 李宝峰 冯乐乐 BAI Hongquan;LV Heping;LI Baofeng;FENG Lele(China Energy Shuohuang Railway Development Co.,Ltd.,Yuanping 034100,China;China Energy Railway Equipment Company Limited Cangzhou Branch,Cangzhou 061100,China)

机构地区：[1]国能朔黄铁路发展有限责任公司,山西原平034100 [2]国能铁路装备有限责任公司沧州分公司,河北沧州061100

出　　处：《电子设计工程》2024年第17期40-44,共5页Electronic Design Engineering

基　　金：国能朔黄铁路公司科技创新项目(SHYP-22-05);国能铁路装备有限责任公司科技项目(TZKY-21-04)。

摘　　要：基于提高铁路网络的可靠性和安全性,同时降低成本的目的,采用了基于深度强化学习(DRL)的铁路维护和更新规划方法,通过考虑成本效益和风险降低,在规划周期内优化更新和维护规划,考虑预测性维护和基于条件的维护任务,并将时间、资源和相关工程约束纳入模型中,以捕捉问题的实际特征,采用双重深度Q网络(DDQN)来克服环境的不确定性,应用优先回放记忆来改善反馈。提出的DDQN方法被应用于一家货运铁路网络,分析结果表明,所提出的方法可以制定出一种最优策略,不仅可以减少预算消耗,还可以提高网络的可靠性和安全性。To enhance the reliability and safety of railway networks while reducing costs,this paper proposes a railway maintenance and update planning approach based on Deep Reinforcement Learning(DRL).The approach optimizes the planning of updates and maintenance within a planning cycle by considering cost⁃effectiveness and risk reduction.It takes into account predictive maintenance and condition⁃based maintenance tasks,as well as incorporating time,resources,and relevant engineering constraints into the model to capture the practical features of the problem.The Dual Deep Q⁃Network(DDQN)is employed to overcome environmental uncertainties,and prioritized replay memory is applied to improve feedback.The proposed DDQN method is applied to a freight rail network.The analysis results indicate that the proposed method can formulate an optimal strategy that not only reduces budget consumption but also improves the reliability and safety of the network.

关 键 词：铁路更新 铁路维护 深度强化学习 双重深度Q网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]