The invention provides a traffic prediction method based on an attention reconstruction neural network, and belongs to the technical field of traffic prediction. The problems of incomplete feature extraction and low model precision in traffic prediction based on a deep learning model at present are solved. Comprising the following steps: acquiring traffic information on a road network through a sensor, and dividing the acquired traffic information on the road network into a nearest time sequence and a periodic time sequence of past several weeks by taking a week as a unit according to a time sequence; a graph attention network GAT based on multi-head attention is adopted to respectively capture a spatial dependency relationship of traffic information between road sections of the latest time sequence and the periodic time sequence; time dependence output of the nearest time sequence and the periodic time sequence is obtained through long and short term feature extraction GRU, dynamic importance between the nearest time sequence and the periodic time sequence is captured through an attention layer, and finally the predicted vehicle speed measured by a sensor at the next moment is output; the method is applied to traffic prediction.

    本发明提供了一种基于注意力重构神经网络的交通预测方法,属于交通预测技术领域;解决了目前基于深度学习模型对交通预测中存在的特征提取不全面、模型精度低的问题;包括如下步骤:通过传感器获取道路网络上的交通信息,并将获取的道路网络上的交通信息根据时序划分为最近的时间序列和以周为单位的过去几周的周期时间序列;采用基于多头注意力的图注意力网络GAT分别捕捉最近的时间序列和周期时间序列的路段之间交通信息的空间依赖关系;分别通过长短期特征提取GRU得到最近的时间序列和周期时间序列的时间依赖性输出,最后通过注意力层来捕捉它们之间的动态重要性,最终输出预测的下一时刻的传感器测得的车辆速度;本发明应用于交通预测。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Traffic prediction method based on attention reconstruction neural network


    Weitere Titelangaben:

    基于注意力重构神经网络的交通预测方法


    Beteiligte:
    CAI JIANGHUI (Autor:in) / YANG HAIFENG (Autor:in) / LIU AIQIN (Autor:in) / XUN YALING (Autor:in) / WANG YUJIA (Autor:in)

    Erscheinungsdatum :

    2023-10-27


    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 / 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



    Dynamic space-time neural network traffic flow prediction method based on attention mechanism

    MENG XIANGFU / XU RUIHANG / FAN HONGYU et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    Graph neural network traffic flow prediction method and system based on attention mechanism

    YU LONGFEI / PENG ZHAOHUI | Europäisches Patentamt | 2020

    Freier Zugriff

    An effective spatial-temporal attention based neural network for traffic flow prediction

    Do, Loan N.N. / Vu, Hai L. / Vo, Bao Q. et al. | Elsevier | 2019


    Traffic flow prediction method based on time-space gated attention map neural network

    XU JIE / GENG ZILI / WU RONGSEN et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Traffic flow prediction method and device based on space-time attention graph neural network

    HUANG WENBING / YUAN JIRUI / TIAN CHUJIE et al. | Europäisches Patentamt | 2024

    Freier Zugriff