The invention relates to the technical field of traffic prediction, and discloses a digraph-based regional traffic prediction method, device and equipment and a medium, which can detect spatio-temporal data of a plurality of target traffic parameters, determine node data of corresponding nodes in a preset node set, generate corresponding digraph data and input the digraph data into a graph neural network for spatio-temporal prediction. And obtaining time sequence prediction data of each node. The digraph data generated according to the node data of each node is generated under the condition that the influence of the node data of any node on the node data of other nodes is effectively considered, and the asymmetric influence between the data of two nodes is effectively considered. According to the method, the directed graph data is input into the graph neural network for space-time prediction, the space-time prediction can be carried out under the condition that the asymmetric influence between different node data is effectively considered, that is, the asymmetric influence between the space-time data of different traffic parameters is effectively considered, and the space-time prediction accuracy of the traffic parameters can be effectively improved.
本发明涉及交通预测技术领域,公开了基于有向图的区域交通预测方法、装置、设备及介质,可以检测出多个目标交通参数的时空数据,确定预设节点集中相应节点的节点数据,生成相应的有向图数据并输入至图神经网络中进行时空预测,得到各节点的时序预测数据。本发明根据各节点的节点数据所生成的有向图数据,是在有效考虑到任一节点的节点数据对其他各节点的节点数据的影响,并且是有效考虑到两个节点数据之间的非对称性影响下生成的。本发明将有向图数据输入至图神经网络进行时空预测,可以在有效考虑到不同节点数据之间的非对称性影响,即有效考虑到不同交通参数的时空数据之间的非对称影响下进行时空预测,能有效提升交通参数的时空预测准确率。
Digraph-based regional traffic prediction method, apparatus and device, and medium
一种基于有向图的区域交通预测方法、装置、设备及介质
2023-11-14
Patent
Electronic Resource
Chinese
European Patent Office | 2021
|Research on Flight Delay Based on Adaptive Agent Digraph
Trans Tech Publications | 2014
|