摘要
【目的】重载列车因线路复杂及纵向力波动大,容易发生安全事故。综合纵向动力学模型与强化学习算法,提出基于强化学习的重载列车操纵优化算法,以降低纵向力波动并提升运行安全性和效率。【方法】通过考虑纵向力约束,将其抽象为强化学习问题,并定义状态空间包括速度、位置、时间、累计制动缓解时间和前两阶段控制力,动作空间包括列车控制力与空气制动决策。以提升运输效率与降低能耗为目标,并确保列车纵向力与运行安全符合要求,基于Q学习算法,为重型货运列车设计了一种智能操纵策略优化方法。【结果】仿真验证表明,提出的纵向力评估模型有效;在长大下坡段优化中,有纵向力约束的优化策略显著降低了最大拉钩力,同时保持较高的运输效率。【结论】基于强化学习的重载列车操纵优化算法,在满足纵向力约束的前提下实现了列车的安全、高效运行。
[Purposes]Heavy-haul trains are prone to safety accidents due to complex railway lines and significant fluctuations in longitudinal forces.By combining a longitudinal dynamics model with reinforcement learning algorithms,this paper proposes a reinforcement learning-based operation optimization algorithm for heavy-haul trains to reduce longitudinal force fluctuations and improve operational safety and efficiency.[Methods]The longitudinal force constraints were considered and abstracted as a reinforcement learning problem.The state space was defined to include speed,position,time,accumulated brake release time,and the control forces from the previous two stages.The action space included train control force and air brake decisions.Aiming to improve transportation efficiency and reduce energy consumption while ensuring longitudinal forces and operational safety meet requirements,a Qlearning algorithm was employed to design an intelligent operation strategy optimization method for heavy-haul trains.[Findings]Simulation results demonstrate the effectiveness of the proposed longitudinal force evaluation model.In the optimization of long downhill sections,the optimized strategy with longitudinal force constraints significantly reduced the maximum coupler force while maintaining high transportation efficiency.[Conclusions]The reinforcement learning-based operation optimization algorithm for heavy-haul trains proposed in this paper achieves safe and efficient train operation under longitudinal force constraints.
作者
李圆
豆飞
姚星
曹振翔
李天祥
LI Yuan;DOU Fei;YAO Xing;CAO Zhengxiang;LI Tianxiang(Guoneng Shuohuang Railway Development Co.,Ltd.,Beijing 100080,China;Southwest Jiaotong University,Chengdu 610000,China)
出处
《河南科技》
2025年第22期34-43,共10页
Henan Science and Technology
关键词
重载列车
列车操纵优化
纵向力
强化学习
列车车钩力
heavy-haul train
train operation optimization
longitudinal force
reinforcement learning
train coupler force
