The invention discloses an urban rail train energy control method based on deep reinforcement learning. The method comprises the following steps that S1, a plurality of train starting samples on an urban rail train platform are collected; s2, extracting train starting sample features and adding labels, and constructing a train starting training data set; s3, training a machine learning model by using the train starting training data set until a test error reaches a preset value or reaches a preset maximum number of training times, and obtaining a trained train starting time prediction model; and S4, inputting the characteristics of a to-be-detected pull-in train into the trained train starting time prediction model, and judging the starting time of the to-be-detected pull-in train. According to the method, the overlapping time of the traction stage of the to-be-started train and the braking stage of the adjacent pull-in train can be longest, so that the purpose of fully utilizing the braking energy to pull the train is achieved.

    本发明公开了一种深度强化学习的城轨列车能量控制方法,包括以下步骤:步骤S1:采集城轨列车平台上若干列车启动样本;步骤S2:提取列车启动样本特征并添加标签,构造列车启动训练数据集;步骤S3:利用所述列车启动训练数据集对机器学习模型进行训练,直到测试误差达到预设值或者到达预设最大训练次数,得到训练后的列车启动时间预测模型;步骤S4:将待测进站列车的特征输入训练后的列车启动时间预测模型,判断待测进站列车的启动时间。本发明可以使待启动列车牵引阶段和相邻进站列车制动阶段的重叠时间最长,从而达到充分利用制动能量牵引列车的目的。


    Access

    Download


    Export, share and cite



    Title :

    Urban rail train energy control method based on deep reinforcement learning


    Additional title:

    一种深度强化学习的城轨列车能量控制方法


    Contributors:
    WANG QIONG (author) / WANG XIAOKAN (author) / CHEN JIAN (author) / XUE JIAO (author) / WU DEQI (author) / ZHANG CHI (author) / WANG XIANGBING (author) / JIANG ZHUPENG (author) / AN HONGHU (author)

    Publication date :

    2024-01-19


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    B61L Leiten des Eisenbahnverkehrs , GUIDING RAILWAY TRAFFIC



    Urban rail train energy-saving operation strategy online optimization method based on reinforcement learning

    WANG XIAOKAN / JIANG KEZHAO / CHEN JIAN et al. | European Patent Office | 2025

    Free access

    Autonomous Train Dispatch Method for Regional Rail Transit Based on Deep Reinforcement Learning

    Wei, Wei / Liu, Ling / Liu, Jun et al. | British Library Conference Proceedings | 2022