The invention discloses a dynamic reinforcement learning enabled heavy-load train group cooperative control method and system, a terminal and a medium. The method comprises the following steps: acquiring real-time operation information of trains in a heavy-load train group; designing a train group cooperative control law according to the deviation between the real-time speed and the reference speed of the train, the speed deviation of the front and rear trains and the distance between the front and rear trains; a reinforcement learning model is established for each train in a train group and is used for dynamically adjusting the weight of a speed deviation item and a distance deviation item in a corresponding train control law, so that the system can learn and adapt to different operation environments in training, and a reward function considers control targets of speed coordination and safe distance, so that the system is more reliable. A penalty function is introduced to process the condition of deviating from an expected state and a safety boundary; and iteratively training each depth determination strategy gradient model to obtain a control law corresponding to the train and act on a traction braking system of the train, so that the train group can be intelligently and cooperatively controlled during high-speed operation, and the operation efficiency and safety are both considered.
本发明公开了一种动态强化学习使能的重载列车群组协同控制方法、系统、终端及介质,其中方法:获取重载列车群组中列车的实时运行信息;根据列车的实时速度与参考速度的偏差、前后车速度偏差以及前后车距离,设计列车群组协同控制律;针对列车群组各列车分别建立强化学习模型,用于动态调整对应列车控制律中速度偏差项和距离偏差项的权重,使得系统能够在训练中学习并适应不同的运行环境,且奖励函数考虑速度协同和安全距离的控制目标,并引入罚函数来处理偏离期望状态及安全边界的情况;迭代训练各深度确定策略梯度模型,获得对应列车的控制律并作用于列车的牵引力制动系统,使得列车群组在高速运行中能够智能协同控制,兼顾运行效率和安全性。
Dynamic reinforcement learning enabled heavy haul train group cooperative control method and system, terminal and medium
动态强化学习使能的重载列车群组协同控制方法、系统、终端及介质
2024-05-31
Patent
Electronic Resource
Chinese
IPC: | B61L Leiten des Eisenbahnverkehrs , GUIDING RAILWAY TRAFFIC |
Parking control method, device and system for heavy haul train, terminal and medium
European Patent Office | 2024
|Method and system for controlling heavy-haul train based on reinforcement learning
European Patent Office | 2021
|On Low Dynamic Actions in Heavy Haul Train
British Library Conference Proceedings
|Heavy haul train braking method, device, system and storage medium
European Patent Office | 2020
|