A New Spatio-Temporal Graph Convolution Network For Traffic Prediction

Accurate traffic flow prediction plays a very important role in the intelligent transportation system (ITS), which is of great importance to traffic planning and traffic management. The traffic flow in the urban road network is influenced by the topology structure, thus the current traffic condition is not only related to the traffic condition of the last few moments, but also influenced by the condition of their neighboring roads. Traditional urban traffic flow prediction methods does not take full account of spatial structure factor, resulting in low prediction accuracy. In order to achieve more accurate prediction result, the prediction model uses graph neural network to extract the road spatial structure, and unlike the previous traffic flow prediction model, a linear layer is used to capture the temporal correlation in the traffic flow. Experiments prove that the model has great prediction ability and better feature extraction capability.