The invention discloses a traffic flow prediction method based on a multi-view dynamic graph convolutional network, and the method comprises the following steps: constructing a relation matrix, and obtaining the spatial dependence characteristics of a traffic flow based on the relation matrix; constructing a prediction model, and inputting the spatial dependency features into the prediction model for fusion processing to obtain spatial-temporal features of different visual angles; and based on a dynamic fusion module, fusing the spatial-temporal characteristics of the different visual angles, and predicting the traffic flow. According to the method, a multi-view coding-decoding end network is utilized, relation matrixes corresponding to different traffic flow view angles are obtained through dynamic training and learning of three branches, more spatial dependence characteristics between stations can be captured, and the final prediction precision is improved.
本发明公开了一种基于多视角动态图卷积网络的交通流预测方法,包括以下步骤:构建关系矩阵,基于所述关系矩阵获取交通流的空间依赖特征;构建预测模型,将所述空间依赖特征输入所述预测模型中进行融合处理,获得不同视角的时空特征;基于动态融合模块,将所述不同视角的时空特征相融合,对交通流进行预测。本发明利用多视角编码‑解码端网络,通过三个分支的动态训练,学习得到不同交通流视角对应的关系矩阵,能捕捉到站点间更多的空间依赖特征,以提高最终的预测精度。
Traffic flow prediction method based on multi-view dynamic graph convolutional network
一种基于多视角动态图卷积网络的交通流预测方法
2022-11-08
Patent
Elektronische Ressource
Chinesisch
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 |
Europäisches Patentamt | 2023
|Europäisches Patentamt | 2022
|Multi-scale traffic flow prediction method based on graph convolutional neural network
Europäisches Patentamt | 2021
|Europäisches Patentamt | 2024
|