Prediction And Implementation Of Short-term Traffic Based On The Combination Model Of SVM And BP Neural Networks

With the progress of science and the development of society,the traffic now makes us feel both convenient and comfortable.However,due to the exponential growth of the number of cars and the blind choice of traffic routes lead to the traffic congestion.This paper is based on the algorithm of intelligent theory to predict the short term traffic flow,and provides a reasonable route for users to avoid traffic congestion.At present,the short-term traffic flow forecasting mainly considers the following factors: Firstly,how to preprocess the traffic flow data,ensure data quality;Secondly,how to use the preprocessed data set to make the effective prediction model;Thirdly,how to choice the suitable prediction algorithm for traffic flow forecasting.Firstly,the original traffic flow data is preprocessed in this paper.The quality of the original traffic flow data,which is used to predict,has an important impact on the prediction results,but because of the traffic flow data acquisition equipment failure and other reasons,the collected data will be noisy and data missing.Therefore,the wavelet analysis method is used to denoise the original traffic flow data,and the data quality in the system is ensured.Secondly,we propose a new prediction model and algorithm for short-term traffic flow prediction.Short term traffic flow prediction plays a very important role in the application of traffic control.Short-term traffic flow has the characteristics of nonlinearity,long time span,uncertainty and instability.Single model has been analysised.In the case of fewer data sets,support vector machine(SVM)has better prediction effect.But generally,the traffic flow data sets are large,and SVM can't cope with large data sets.In this paper,a prediction algorithm based on SVM(support vector machine)and BP neural network is proposed in this paper.The optimal weighting rule(Optimal weighted rule)is used to optimize our prediction algorithm.Through the simulation experiment,from the mean absolute percentage error and the root mean square error,our combined algorithm is superior to the SVM and BP neural network algorithms.Finally,the traffic flow prediction system is designed and implemented.Based on the demand analysis of traffic flow prediction system,we designed the overall framework of the system and the design of the system database.Through the compilation of server side and client side code,we have realized the short-term traffic flow prediction system of SVM and BP neural network combined model.