The invention provides a training method of a traffic prediction model, a traffic prediction method and related equipment. The training method comprises the steps of obtaining a time sequence traffic volume sequence of a plurality of traffic nodes and a corresponding hypergraph structure based on an input layer; performing one-dimensional expansion convolution processing on the time sequence traffic volume sequence through a time sequence feature extraction layer, and adjusting a second convolution feature according to a nonlinear conversion value domain of a first convolution feature to obtain a time sequence feature corresponding to each traffic node; performing feature aggregation weighting on the intra-edge neighborhood transfer message and the inter-edge neighborhood transfer message corresponding to each hypergraph structure based on a graph message transfer layer to obtain a spatial feature corresponding to each traffic node; and a traffic prediction result is obtained based on the full-connection output layer according to the time sequence features and the spatial features, and the model is updated according to the model loss to obtain a traffic prediction model, so that a smooth phenomenon caused by frequency domain graph convolution operation is avoided, and the prediction precision of node-level traffic prediction can be effectively improved.
本申请提供了一种交通预测模型的训练方法、交通预测方法以及相关设备,训练方法包括:基于输入层得到多个交通节点的时序交通量序列和对应的超图结构;通过时序特征提取层对时序交通量序列进行一维膨胀卷积处理,并根据第一卷积特征的非线性转换值域对第二卷积特征进行调节,得到每个交通节点对应的时序特征;基于图消息传递层分别对每个超图结构对应的边内邻域传递消息和边间邻域传递消息进行特征聚合加权,得到每个交通节点对应的空间特征;基于全连接输出层根据时序特征和空间特征得到交通预测结果,并根据模型损失进行模型更新得到交通预测模型,以避免频域图卷积操作所导致的平滑现象,进而可以有效提高进行节点级的交通预测的预测精度。
Training method of traffic prediction model, traffic prediction method and related equipment
交通预测模型的训练方法、交通预测方法以及相关设备
2025-02-18
Patent
Electronic Resource
Chinese
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 |
Traffic time prediction model training method, traffic time prediction method, and equipment
European Patent Office | 2021
|Traffic track prediction model training method, traffic track prediction method and related device
European Patent Office | 2024
|European Patent Office | 2024
|Traffic flow prediction model training method, traffic flow prediction method, device and equipment
European Patent Office | 2023
|European Patent Office | 2024
|