The invention discloses a trunk line dynamic coordination training method based on deep reinforcement learning, and the method comprises the steps: obtaining the current traffic state data of a target intersection and an adjacent intersection from a preset simulation environment, inputting the current traffic state data into a Double DQN model, and obtaining a plurality of candidate phase motions and a target phase value; selecting a pre-execution phase action by using a preset agent, and determining a shared reward value and an experience sample of the pre-execution phase action based on the target reward function; storing the generated experience samples into a shared experience pool; determining a plurality of target experience samples from the shared experience pool; and training the estimation network of the Double DQN model by using a target experience sample to obtain a target trunk line dynamic coordination model. By using the method, the traffic change of the intersection can be accurately reflected. The invention further discloses a main line dynamic coordination control method. By using the method, the coordination control of the arterial traffic signal is realized.

    本发明公开了一种基于深度强化学习的干线动态协调训练方法,包括从预设仿真环境中获取目标交叉路口及其相邻交叉路口的当前交通状态数据,将当前交通状态数据输入DoubleDQN模型获得多个候选相位动作和目标相位价值;利用预设智能体选取预执行相位动作,并基于目标奖励函数确定预执行相位动作的共享奖励值和经验样本;将生成的经验样本存入共享经验池中;从所述共享经验池中确定多个目标经验样本;利用目标经验样本对所述DoubleDQN模型的估计网络进行训练,以获得目标干线动态协调模型。利用该方法能够准确的反应交叉路口的交通变化。本发明还公开了干线动态协调控制方法。利用该方法实现了干线交通信号的协调控制。


    Access

    Download


    Export, share and cite



    Title :

    Trunk line dynamic coordination training method and trunk line dynamic coordination control method based on deep reinforcement learning


    Additional title:

    基于深度强化学习的干线动态协调训练方法及干线动态协调控制方法


    Contributors:
    YE BAOLIN (author) / LIU ZHIMIN (author) / ZHU YAODONG (author) / WANG XIANG (author) / CHEN BIN (author) / LU YIXIA (author)

    Publication date :

    2023-06-23


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Full-induction type trunk line green wave coordination control method

    CHEN BIN / LIU HONG / WANG XIAOLIN et al. | European Patent Office | 2023

    Free access

    Coordination control method and device based on trunk line multi-period automatic division

    YAN JUN / HUANG LEI | European Patent Office | 2024

    Free access

    Tidal traffic state trunk line signal coordination design method and device

    WANG HAO / PENG XIANYUE | European Patent Office | 2020

    Free access

    Trunk line coordination control method considering road running speed and intersection traffic state

    REN GANG / ZHOU PINGSHAN / LI HAOJIE et al. | European Patent Office | 2021

    Free access

    Trunk line coordination control method, device and equipment for road intersection group and medium

    ZHANG JUN / CHEN JUNDE / ZHANG QING et al. | European Patent Office | 2023

    Free access