Urban road speed prediction method considering spatial-temporal characteristics of traffic network

The invention discloses an urban road speed prediction method considering spatial-temporal characteristics of a traffic network. The method is used for predicting traffic flow of urban road sections. The method comprises the following steps of: firstly, collecting traffic road network data information, such as speed time sequence of road sections, an adjacency matrix for describing an adjacency relation between the road sections and traffic events occurring in each road section, analyzing and modeling the collected data, and dividing the data into a training set and a test set; secondly, building a DG-GRU model by using a graph attention network, a gating circulation unit and a dense connection network, and extracting spatial-temporal characteristics of a traffic road network; and finally, the training set is sent to a DG-GRU model for training, model parameters are adjusted, so that the model achieves the optimal prediction performance, and the performance of the model is evaluated by using the test set. The