Research on High Speed Vehicle Density Prediction Based on Combined Model

Abstract

In this study, the vehicle density problem is modeled mathematically, aiming at the demand for accurate prediction of vehicle density at popular high-speed toll stations.The model is transformed into a time series processing and prediction problem, and the classical auto regression moving average model (ARIMA) and gray theory model (GM) are introduced to model the traffic flow series.Considering the difference in processing ability of GM and ARIMA for time series with different time granularity, the two methods are integrated, which used the radial basis function network to correct the residual of prediction results. The network uses RBF for information processing, which has better approximation effect for nonlinear and discontinuous changes and significantly improves the prediction accuracy of the algorithm.