The invention provides a train temporary parking strategy optimization method based on deep reinforcement learning. The train temporary parking strategy optimization method is characterized in that an environment module, a knowledge module, an experience playback pool and a plurality of agents are included; the optimization method comprises the following steps: the knowledge module distributes a temporary stop instruction sequence to each agent; each agent interacts with the environment module by adopting an Actor-Critic algorithm to obtain a temporary stop time sequence and an experience array; empirical arrays obtained by the intelligent agents are all put into an empirical playback pool; and after the experience playback pool is fully loaded, each agent randomly extracts an experience array from the experience playback pool to update the corresponding strategy network parameters and value network parameters. By the adoption of the method, the optimization speed and quality of the train temporary stop strategy are improved, and then the train temporary stop accuracy and riding comfort are improved.
本发明提供了一种基于深度强化学习的列车临时停车策略的优化方法,其特征在于:包括环境模块、知识模块、经验回放池和多个智能体;所述优化方法包括:所述知识模块将临停指令序列分发给各个智能体;各个智能体均采用Actor‑Critic算法与环境模块交互以获取临停时间序列和经验数组;将各个智能体获取的经验数组均放入经验回放池;当经验回放池满载后,各个智能体均从经验回放池中随机抽取经验数组对各自对应的策略网络参数和价值网络参数进行更新。采用本发明的方法提高了列车临停策略的优化速度和质量,进而提高了列车临停的准确度和乘坐舒适性。
Train temporary parking strategy optimization method based on deep reinforcement learning
基于深度强化学习的列车临时停车策略的优化方法
2023-08-18
Patent
Electronic Resource
Chinese
IPC: | G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06Q Datenverarbeitungssysteme oder -verfahren, besonders angepasst an verwaltungstechnische, geschäftliche, finanzielle oder betriebswirtschaftliche Zwecke, sowie an geschäftsbezogene Überwachungs- oder Voraussagezwecke , DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Train accurate parking control method based on deep reinforcement learning
European Patent Office | 2022
|Deep reinforcement learning method suitable for automatic parking of high-speed train
European Patent Office | 2022
|Train Regenerative Braking Strategy Optimization Based on Reinforcement Learning
Springer Verlag | 2023
|Train operation scheduling optimization method based on deep reinforcement learning
European Patent Office | 2023
|Automatic parking method based on deep reinforcement learning
European Patent Office | 2020
|