Highway traffic flow forecasting based on spatio-temporal relationship

With the rapid development of economy and science and technology, the problem of highway congestion is gradually serious. It is of great significance for the construction of intelligent highways to forecast the traffic flow of highways, so as to carry out traffic congestion warning and diversion induction in advance in order to improve the efficiency of highway traffic. In addition, traffic flow has complex spatio-temporal dependencies, and most of the existing methods capture the temporal and spatial dependencies individually, making it difficult to model these two dependencies simultaneously. To address these two issues, a highway traffic flow forecasting method based on dynamic graph convolutional network with spatio-temporal feature extraction module is proposed. First, through the dynamic graph