The invention discloses a traffic flow prediction method based on a dynamic multi-graph space-time synchronization network. The method comprises the following steps: generating traffic flow data of a time sequence; designing a feature enhancement module, and fusing the traffic flow data with external factors; defining multiple diagrams in time and space, and constructing two dynamic time-space synchronization diagrams; a dynamic multi-graph space-time synchronization network is constructed, four dynamic multi-graph space-time synchronization layers are overlaid on the network, and two parallel dynamic multi-graph space-time synchronization modules are deployed in each dynamic multi-graph space-time synchronization layer, so that space-time dependence of traffic flow data is synchronously modeled. According to the method, the traffic flow data and external factors are fused, the influence of weather, temperature and other meteorological factors on the traffic flow is considered, and meanwhile, different dynamic space-time synchronization diagrams are constructed, so that the space-time correlation of the traffic flow data is synchronously extracted from multiple aspects and multiple angles, and then the model achieves accurate prediction precision.
本发明公开了一种基于动态多图时空同步网络的交通流量预测方法,包括:生成时间序列的交通流数据;设计特征增强模块,将交通流数据与外部因素融合;定义时间和空间上的多图,并构造两种动态时空同步图;构建动态多图时空同步网络,该网络叠加四层动态多图时空同步层,每层动态多图时空同步层中部署两种平行的动态多图时空同步模块,以此同步建模交通流数据的时空依赖。本发明将交通流数据与外部因素融合,考虑了天气、温度等气象因素对交通流的影响,同时构造了不同的动态时空同步图,以从多方面、多角度同步提取交通流数据的时空相关性,进而使模型达到精确的预测精度。
Traffic flow prediction method based on dynamic multi-graph space-time synchronization network
一种基于动态多图时空同步网络的交通流量预测方法
2024-11-22
Patent
Electronic Resource
Chinese
European Patent Office | 2024
|European Patent Office | 2023
|Space-time adaptive dynamic graph convolutional network traffic flow prediction method
European Patent Office | 2024
|European Patent Office | 2025
|European Patent Office | 2023
|