Traffic Flow Prediction of Expressway Section Based on RBF Neural Network Model

In order to further improve the prediction accuracy of expressway traffic flow, this study proposed an RBF neural network model. Firstly, RBF is used to train the model by using ETC (Electronic Toll Collection) gantry historical data considering the time-varying characteristics of the flow, to ensure the similarity of the flow curve and robustness of the model. Then, taking three typical ETC gantries from thirty gantries of Beijing-Shanghai Expressway in Shandong province as an example, the accuracy of the model is verified by using the historical operation data of them during holidays. The results show that: (1) The flow of ETC gantry section in holidays predicted by RBF is closer to the actual value, and the prediction accuracy is significantly better than that of BP and ELMAN. (2) The MAE is within 75 veh/min, the RMSE is within 6veh/min, and the MAPE is less than 4.5%.