The invention discloses a local traffic optimization method based on reinforcement learning and a generative adversarial network, and the method comprises the steps: building a training model, employing a generative adversarial network to autonomously improve the accuracy of the model, and predicting the traffic flow data at a specified moment through training the real traffic flow data detected at a certain intersection; the real traffic flow data and the virtual traffic flow data are trained and output by adopting Q learning to form a Q value table, an optimal local traffic optimization strategy is obtained by adopting a reward function, and the traffic signal lamp period adjustment efficiency is greatly improved by utilizing the advantage of interactive learning of reinforcement learning, so that the traffic signal lamp period adjustment efficiency is improved; and whether the congestion condition is relieved or not is verified by adjusting the current congestion level and the traffic light time ratio of a certain intersection, the optimal traffic light time ratio is obtained through repeated and continuous optimization, finite-time optimal training of Q learning is achieved through the heuristic self-game thought of the generative adversarial network, local traffic optimization is achieved, and finally, an optimal adjustment scheme is obtained, so that the local traffic optimization capability is improved.

    一种基于强化学习与生成式对抗网络的局部交通优化方法,包括建立训练模型,采用生成对抗网络自主提升模型的准确率,通过训练某个路口检测到的真实车流量数据预测指定时刻的车流量数据;采用Q学习对真实车流量数据和虚拟车流量数据进行训练输出动作形成Q值表,采用奖赏函数,得到最佳局部交通优化策略,利用强化学习交互式学习的优点,大大提升了交通信号灯周期调整的效率,由某个路口的当前拥堵级别和红绿灯信号灯时间比调整来验证拥堵情况是否有所缓解,以此往复不断优化得到最佳的红绿灯时间比,再利用生成式对抗网络的启发自博弈思想实现对Q学习的有限时间最佳训练,实现局部交通优化,最终得到最优调整方案,从而提升局部交通优化能力。


    Access

    Download


    Export, share and cite



    Title :

    Local traffic optimization method based on reinforcement learning and generative adversarial network


    Additional title:

    基于强化学习与生成式对抗网络的局部交通优化方法


    Contributors:
    LIU XINCHENG (author) / XUAN FAN (author) / XIAO TONG (author) / XU CUI (author) / ZHOU GUODONG (author)

    Publication date :

    2021-06-18


    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



    Traffic flow prediction method based on generative adversarial network

    XU DONGWEI / LIN ZHENQIAN / YANG HAO et al. | European Patent Office | 2021

    Free access

    Generative Adversarial Network-Based Regional Epitaxial Traffic Flow Prediction

    Kang, Yan / Li, Jinyuan / Lee, Shin-Jye et al. | Springer Verlag | 2019


    Urban traffic flow prediction method based on generative adversarial network

    ZHU NINGBO / PU BIN / FAN XINXIN et al. | European Patent Office | 2020

    Free access

    Missing traffic state data filling method based on generative adversarial network

    LI XU / ZUO JINPENG / SHI TIANJING et al. | European Patent Office | 2023

    Free access

    Endo-atmospheric maneuver penetration strategy based on generative adversarial reinforcement learning

    HUI, Yaoluo / LI, Xiumin / LIANG, Chen et al. | Elsevier | 2025

    Free access