The reasonable control strategy of urban traffic signal can alleviate the congestion at intersections. To improve the learning efficiency and stability of intelligent signal control algorithms, this paper proposes an urban traffic signal control strategy based on the Rainbow algorithm. Based on the classical DQN algorithm, the Rainbow algorithm introduces the Double Q-Network, Noisy Network, Dueling-DQN, and Distribution-DQN into the network structure, and introduces the prioritized experience replay technology and multi-step learning into the algorithm training process. Taking the intersections of Niangre Road and Dangre Road in the Lhasa City as an example, three simulation scenarios are set based on the traffic flow intensity. Three traffic signal control strategies are conducted by using fixed-time signal control algorithm, classical DQN algorithm, and Rainbow algorithm, respectively. Results demonstrate that the Rainbow algorithm can converge quickly and reduce vehicle delay time, queue length effectively.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Intelligent Traffic Signal Control Strategy Based on Rainbow Algorithm


    Contributors:
    MA, Die (author) / OU, Jushang (author) / JU, Yanni (author)


    Publication date :

    2023-10-28


    Size :

    2588982 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Research on Strategy Intelligent Control of Traffic Signal

    Zhang, Yan Min ;Luo, Hai Bing ;Wang, Jian Qiang | Trans Tech Publications | 2015



    Intelligent traffic intelligent signal control method

    PEI CHUNHONG / FENG QIANG | European Patent Office | 2023

    Free access

    Intelligent traffic control signal lamp

    LIAO XIN / XU LONG / LIU CHUANGQING | European Patent Office | 2024

    Free access

    Intelligent traffic signal control method

    LIU PENG / ZHANG ZHEN / CAO XUDONG et al. | European Patent Office | 2021

    Free access