The invention relates to an expressway traffic flow prediction method based on Transform and a graph attention network. A traffic flow prediction model comprises the graph attention network and the Transform network. The graph attention network is used for capturing the spatial correlation of the traffic flow, the Transform is used for capturing the temporal correlation of the traffic flow, and a prediction result of each step is output by using a linear layer. According to the model, traffic flow is divided into recent flow, daily cycle flow and weekly cycle flow, and periodical explicit modeling is carried out on the traffic flow. In addition, the model introduces weather factors into the predictive model for dynamically adjusting the degree of dependency between different time slices. In the graph attention network, an attention mechanism is combined with a predefined adjacency matrix to dynamically model spatial correlation of traffic flow. And modeling is carried out on the time correlation of the traffic flow by using a Transform network, so that the precision of long-term prediction is improved. The traffic flow prediction method is based on data driving, artificial feature engineering design is not needed, and implementation is more flexible.

    本发明涉及一种基于Transformer与图注意力网络的高速公路交通流量预测方法,该交通流量预测模型包含图注意力网络与Transformer网络。图注意力网络用于捕获交通流量的空间相关性,Transformer用于捕获交通流量的时间相关性,利用线性层输出每一步的预测结果。该模型将交通流量划分为近期流量、日周期流量、周周期流量,对交通流量的周期性显式建模。此外,该模型将天气因素引入该预测模型中用于动态地调整不同时间片之间的依赖程度。图注意力网络中使用注意力机制结合预定义邻接矩阵对交通流量的空间相关性动态建模。使用Transformer网络对交通流量的时间相关性进行建模,提高长期预测的精度。本发明中涉及的交通流量预测方法基于数据驱动,无需人工特征工程设计,实施起来更加灵活。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Highway traffic flow prediction method based on Transform and graph attention network


    Weitere Titelangaben:

    一种基于Transformer与图注意力网络的高速公路交通流量预测方法


    Beteiligte:
    JIANG CONG (Autor:in) / SONG YUN (Autor:in) / DENG ZELIN (Autor:in)

    Erscheinungsdatum :

    09.05.2023


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung



    Traffic flow prediction based on graph attention network

    Zhu, Wenyan / Kong, Hoiio / Cai, Wenzheng et al. | SPIE | 2024


    Traffic flow prediction method based on graph attention convolution network

    ZHENG HONG / ZHANG SIKAI / LIU JIAMOU et al. | Europäisches Patentamt | 2020

    Freier Zugriff

    Traffic flow prediction method based on cyclic attention coupled graph convolutional network

    CHEN LING / CHEN WEIQI | Europäisches Patentamt | 2020

    Freier Zugriff

    Traffic flow prediction method based on Transform adaptive adversarial graph neural network

    XIA ZHENCHANG / YUAN LINGTIAN / QIU CHENGYI et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Node adaptive learning graph attention neural network traffic flow prediction method

    LOU JUNGANG / HUANG XUXIANG / SHEN QING et al. | Europäisches Patentamt | 2024

    Freier Zugriff