The invention discloses a mixed multi-ramp cooperative confluence control method based on multi-agent reinforcement learning. For a highway continuous entrance ramp road network where intelligent network connection vehicles and manual driving vehicles are mixed, a multi-agent reinforcement learning algorithm is used for training to obtain a multi-ramp cooperative control model, and traffic flow operation conditions of main lines and ramps near all entrance ramps are monitored in parallel. And the intelligent network connection vehicles on the main line and the ramps are controlled to cooperatively complete the confluence process. At each entrance ramp, when a main line generates traffic jam due to ramp confluence, a roadside controller retrieves the intelligent network connection vehicles in the ramps and enables the intelligent network connection vehicles to stop and wait for giving way after arriving at a confluence point, the parking waiting time is determined by a multi-ramp cooperative control model, and the traffic efficiency of a plurality of entrance ramps is considered at the same time when the model is trained. And for the ramp intelligent network connection vehicles finishing waiting, the roadside controller enables the intelligent network connection vehicles on the outermost main line lane to cooperatively and slowly wait at the upstream of the confluence point.

    本发明公开了一种基于多智能体强化学习的混行多匝道协作合流控制方法。针对智能网联车辆和人工驾驶车辆混行的高速公路连续入口匝道路网,应用多智能体强化学习算法训练得到多匝道协作控制模型,并通过分别并行监测各入口匝道附近主线和匝道的交通流运行状况,控制主线和匝道上的智能网联车辆协作完成合流过程。在各入口匝道处,当主线由于匝道合流产生交通拥堵时,路侧控制器检索匝道中的智能网联车辆,令智能网联车到达合流点后停车等待让行,其中停车等待时间由多匝道协作控制模型确定,模型训练时会同时考虑多个入口匝道交通效率。对结束等待的匝道智能网联车辆,路侧控制器令最外侧主线车道上智能网联车在合流点上游协作慢行等待。


    Access

    Download


    Export, share and cite



    Title :

    Mixed multi-ramp cooperative confluence control method based on multi-agent reinforcement learning


    Additional title:

    一种基于多智能体强化学习的混行多匝道协作合流控制方法


    Contributors:
    YAN JUNQI (author) / YU HONGXIN (author) / ZHANG LIHUI (author) / WANG YIBING (author)

    Publication date :

    2023-04-04


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Ramp confluence cooperative control method and system based on confluence time optimization

    BIAN YOUGANG / WANG XIAOWEI / XIE GUOTAO et al. | European Patent Office | 2020

    Free access

    Vehicle ramp entrance confluence control method based on deep reinforcement learning

    TONG NING / LI FENGQI / WANG KUNRUI | European Patent Office | 2023

    Free access

    Multi-vehicle cooperative control method, device and equipment in ramp confluence area and storage medium

    KONG WEIWEI / CAI TIANMAO / JI PENGXIAO et al. | European Patent Office | 2023

    Free access

    Ramp confluence area mixed traffic flow control method based on time-space diagram neural network reinforcement learning

    XU DONGWEI / QIU QINGWEI / GAO GUANGYAN et al. | European Patent Office | 2024

    Free access

    Multi-intelligent network connection vehicle cooperative confluence control method for high-speed ramp entrance

    ZHUANG WEICHAO / LIU CHANG / YIN GUODONG et al. | European Patent Office | 2020

    Free access