The embodiment of the invention discloses a method for adaptively learning traffic flow prediction under a dynamic traffic condition, and the method comprises the steps: carrying out the dynamic learning of a road topological graph from global and local angles through employing a graph structure, and obtaining the optimal adjacency matrix representation of the road traffic condition; capturing a long-term space-time dependency relationship of the traffic data through a multi-head attention mechanism of a Transform module; extracting a short-term dependency relationship of the time sequence by using a gating time convolutional network; and capturing a short-term spatial dependency relationship by using a graph convolution mechanism. By adopting the method, the adaptive learning and the high dynamic property of the adjacent graph are fully considered, and the adjacent graph learning is carried out from the angles of the global graph and the local graph, so that the short-term spatial change can be better captured.

    本发明实施例公开了一种用于动态交通状况下的自适应学习交通流预测的方法,使用图结构从全局和局部两个角度对道路拓扑图进行动态学习,获取道路交通状况的最佳邻接矩阵表示;通过Transformer模块的多头注意力机制捕捉到交通数据的长期时空依赖关系;使用门控时间卷积网络提取时间序列的短期依赖关系;使用图卷积机制捕获短期的空间依赖关系。采用本发明,充分考虑了邻接图的自适应学习和高度动态性,从全局图形和局部图形角度进行邻接图学习,可以更好地捕捉到短期的空间变化。


    Access

    Download


    Export, share and cite



    Title :

    Adaptive learning traffic flow prediction method under dynamic traffic condition


    Additional title:

    一种用于动态交通状况下的自适应学习交通流预测的方法


    Contributors:
    ZHOU TENG (author) / LI ZEQUAN (author) / YANG LISHA (author)

    Publication date :

    2023-09-08


    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



    TRAFFIC CONDITION PREDICTION SYSTEM AND TRAFFIC CONDITION PREDICTION METHOD

    OBA YOSHIKAZU / UENO HIDEKI | European Patent Office | 2021

    Free access

    Traffic condition prediction system and traffic condition prediction method

    CHEN YEN-LIN / CHEN HSIU-CHIH / YU CHAO-WEI et al. | European Patent Office | 2022

    Free access

    Adaptive traffic signal control method combining traffic flow prediction and reinforcement learning

    PI JIATIAN / YANG XINMIN / WU CHANGZHI | European Patent Office | 2023

    Free access

    TRAFFIC CONDITION PREDICTION DEVICE, AND TRAFFIC CONDITION PREDICTION METHOD

    AOKI YASUHIRO / SUZUKI YOSHIHIKO / NARUSE KOSUKE | European Patent Office | 2021

    Free access