Research On The Aapplication Of Neural Network In Traffic Flow Forecast Of Expressway

Traffic flow analysis and prediction is not only the core content of intelligent transportation system, but also the important basis for monitoring information service, traffic control and guidance.Based on the theory of traffic flow and neural network, this paper mainly discusses two kinds of function models, which are applied to the traffic flow forecasting of expressway, namely BP neural network and wavelet neural network. The main research contents are as follows:1 Through the study and research of traffic flow and neural network, the prediction scheme of the highway traffic flow based on the BP neural network and wavelet neural network is put forward, and the modeling process and forecasting performance of BP network are analyzed. And for the deviation of the standard BP network model is too large and the network can not be convergence, the BP network is improved by using additional momentum and adaptive adjusting learning rate.2 The predictive simulation to the actual traffic flow data of Liaoning expressway section of fuying as the sample is done. Through the comparison of forecast data, the advantages and disadvantages of BP neural network and wavelet neural network are analyzed from the aspects of system stability, learning efficiency and prediction accuracy.3 Overall, the standard BP neural network prediction is less effective, and can not meet the design requirements. However, the improved BP neural network and wavelet neural network model for the prediction of the traffic flow of the freeway is basically in line with the actual traffic demand. In forecasting precision, wavelet network is better than BP neural network; In the aspect of system stability, BP network is better than wavelet network.