The invention discloses a traffic flow prediction method based on a space-time attention mechanism, and the method comprises the steps: carrying out the coding processing of previous traffic data X0, obtaining a data embedded feature vector X, obtaining the time attention TSA, the adjacent space attention DisSSA and the similar space attention SimSSA through self-attention operation, carrying out the fusion of the above features based on a W parameter matrix, and carrying out the prediction of the traffic flow. A fusion feature vector Xw is obtained through a full-connection feedforward network, layer normalization and residual connection processing; and performing convolution on the fusion feature vector Xw to form jump connection processing, converting the fusion feature vector Xw into a jump vector Xh, and performing convolution prediction through the jump vector Xh to obtain predicted traffic flow. Features of previous traffic data are comprehensively extracted by using a space-time embedding mechanism, and complex time and space features can be flexibly captured based on a space-time attention mechanism, so that the traffic flow prediction efficiency is improved. And then a W parameter-based fusion method is utilized to further fuse valuable characteristics, so that the accuracy of traffic flow prediction can be remarkably improved, and the applicability is high.

    本发明公开了一种基于时空注意力机制的交通流量预测方法,包括通过对往期交通数据X0编码处理,获取数据嵌入特征向量X,并通过自注意力操作获取时间注意力TSA、邻接空间注意力DisSSA和相似空间注意力SimSSA,将上述特征以基于W参数矩阵融合,并经过全连接前馈网络、层归一化和残差连接处理获得融合特征向量Xw;对融合特征向量Xw卷积组成跳跃连接处理,转化为跳跃向量Xh,并通过跳跃向量Xh卷积预测,获取预测交通流量本申请通过使用时空嵌入机制全面的提取往期交通数据的特征,并基于时空注意力机制能够灵活捕获复杂时间和空间特征,然后利用基于W参数的融合方法,能够进一步融合到有价值的特性,从而能够显著提高对交通流量预测的准确性,且适用性强。


    Access

    Download


    Export, share and cite



    Title :

    Traffic flow prediction method based on space-time attention mechanism


    Additional title:

    一种基于时空注意力机制的交通流量预测方法


    Contributors:
    WANG JIAYING (author) / YANG HENG (author) / SHAN JING (author) / SONG XIAOXU (author) / NIU LIANQIANG (author)

    Publication date :

    2024-05-03


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06Q Datenverarbeitungssysteme oder -verfahren, besonders angepasst an verwaltungstechnische, geschäftliche, finanzielle oder betriebswirtschaftliche Zwecke, sowie an geschäftsbezogene Überwachungs- oder Voraussagezwecke , DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Traffic flow prediction method based on space-time bidirectional attention mechanism

    ZHOU YIMIN / YANG SHUDONG | European Patent Office | 2024

    Free access

    Traffic flow prediction method based on cyclic space-time attention mechanism

    XU XIAO / ZHANG LEI / LIU BAILONG et al. | European Patent Office | 2024

    Free access

    Traffic flow prediction method based on space-time decomposition and attention mechanism

    CHEN QIAOSONG / XIAO SHUANGWANG / TANG CHANGTIAN et al. | European Patent Office | 2025

    Free access

    Traffic flow prediction method based on space-time attention network

    MA CHUANG / YAN LI / LIU SHUAIWU et al. | European Patent Office | 2023

    Free access

    Traffic prediction method based on multi-scale space-time attention mechanism

    LU ZIBAO / ZHANG JIALI / AN CHEN et al. | European Patent Office | 2025

    Free access