SHORT-TERM TRAFFIC FLOW FORECASTING BASED ON MATHEMATICAL MODELS

In urban traffic flow prediction, the lack of analysis of the spatial correlation characteristics of traffic flow leads to a large deviation between the prediction results and the real traffic flow data. Therefore, the study takes into account the spatial correlation characteristics of traffic flow and introduces the graph convolution algorithm into the mathematical model, so that the coupled mathematical model can meet the requirements of time series processing of traffic flow data. Finally, the validity of the model was verified by testing the MAE and RMSE values of the model.