The invention discloses a traffic flow prediction method and device based on a dynamic ordinary differential graph neural network, and the method employs a single encoder architecture, and system modules comprise a road traffic network semantic matrix building module, a spatial dependence capturing module, a time dependence extraction module, a dynamic graph ordinary differential network module, and a graph convolutional neural network module. A feature fusion gating unit, a traffic flow prediction module and a terminal design module. In order to solve the problem of Over-Smoothing of graph neural network node data in a traditional traffic flow prediction method, a dynamic graph ordinary differential network is introduced, an analytical solution of a road traffic network topology structure is abstracted by using an ordinary differential equation, spatial-temporal features are extracted by constructing a spatial-temporal attention module, and the spatial-temporal features are extracted by using a spatial-temporal attention module. And a feature fusion gating unit is used to further capture road traffic flow spatio-temporal data features, so that the traffic flow prediction precision is improved.
本发明公开了基于动态常微分图神经网络的交通流量预测方法及装置,本发明方法使用单编码器架构,系统模块包括:建立道路交通网络语义矩阵模块,空间依赖捕捉模块,时间依赖提取模块,动态图常微分网络模块,图卷积神经网络模块,特征融合门控单元,交通流量预测模块和终端设计模块。本发明为解决传统交通流量预测方法中图神经网络节点数据相似(Over‑Smoothing)问题,引入动态图常微分网络,使用常微分方程抽象出道路交通网络拓扑结构的解析解,通过构建时空注意力模块提取时空特征,并使用特征融合门控单元进一步捕捉道路交通流时空数据特征,从而提升交通流量预测精度。
Traffic flow prediction method and device based on dynamic ordinary differential graph neural network
基于动态常微分图神经网络的交通流量预测方法及装置
2024-03-29
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / 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 |
Traffic flow prediction method based on dynamic graph neural network
European Patent Office | 2022
|Traffic flow prediction method based on graph neural network
European Patent Office | 2024
|Traffic flow real-time prediction method and device based on dynamic graph neural network
European Patent Office | 2023
|European Patent Office | 2024
|Airport traffic flow prediction method based on graph neural network
European Patent Office | 2023
|