A learning algorithm and forecasting model of the short-term traffic flow based on wavelet neural network

Dynamic analysis and forecasting of traffic flow in intelligent transportation system is of great significance. Based on the time- and frequency-domain features of wavelets, wavelet transform is combined with BP neural network and a method of forecasting short-term traffic flow based on wavelet neural network is proposed along with the specific neural network learning algorithm. Simulation analysis is performed using field data and Matlab software. The results indicate that the forecasting of short-term traffic flow based on wavelet neural network is reliable. Then a comparison is made with conventional BP neural network. It is confirmed that the forecasting model based on wavelet neural network overcomes the non-linear approximation defect with BP neural network. This model is especially suitable for forecasting short-term traffic flow.