The invention relates to a single-intersection intersection traffic light control method based on improved deep reinforcement learning, and the method comprises the steps: obtaining the density of a current vehicle, the speed of the current vehicle and the waiting time of the current vehicle at a controlled single intersection, and defining an intelligent agent state value S based on the density of the current vehicle, the speed of the current vehicle and the waiting time of the current vehicle, an action space A, an action selection strategy epsilon and a reward function r; creating a neural network, a target network and an experience pool E, storing the agent state value S, the action space A, the action selection strategy epsilon and the reward function r in the experience pool E, and sampling a batch from the experience pool E to train the created neural network; updating parameters of the trained neural network to a target network; and based on the trained neural network and the parameter-updated target network, controlling the intersection traffic light of the single intersection. And an experience playback mechanism is introduced, and previous experiences are utilized, so that the utilization rate of samples is improved, the learning speed is accelerated, and the model converges more quickly.

    本发明涉及基于改进的深度强化学习单交叉口路口交通灯控制方法,获取当前被控制单交叉路口的当前车辆的密度、当前车辆速度以及当前车辆等待时间,并基于当前车辆的密度、当前车辆速度以及当前车辆等待时间定义智能体状态值S,动作空间A,动作选择策略ε和奖励函数r;创建神经网络和目标网络和经验池E,将智能体状态值S,动作空间A,动作选择策略ε和奖励函数r存储于经验池E中,并从中采样一个批次对创建的神经网络进行训练;将训练后的神经网络的参数更新至目标网络;基于训练后的神经网络和参数更新的目标网络对单交叉口路口交通灯进行控制。引入经验回放机制利用过往的经验,提高了样本的利用率,加快了学习速度,让模型更快的收敛。


    Access

    Download


    Export, share and cite



    Title :

    Single-intersection intersection traffic light control method based on improved deep reinforcement learning


    Additional title:

    基于改进的深度强化学习单交叉口路口交通灯控制方法


    Contributors:
    LIU BINGYAN (author) / GUO HONG (author) / DU JUNLIN (author) / MA WANLI (author) / TANG WENJIE (author)

    Publication date :

    2024-12-13


    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 / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    Multi-intersection traffic signal control method based on deep reinforcement learning

    LIU LIJUAN / BAI GUANGMING | European Patent Office | 2023

    Free access

    Multi-intersection traffic signal control method based on deep reinforcement learning

    DENG HENG / WANG YULONG / GAO YANG et al. | European Patent Office | 2024

    Free access


    A Deep Reinforcement Learning Agent for Traffic Intersection Control Optimization

    Garg, Deepeka / Chli, Maria / Vogiatzis, George | IEEE | 2019


    Single-intersection signal control method based on deep reinforcement learning algorithm

    HUANG YIWANG / WU QIAN | European Patent Office | 2023

    Free access