Highway Traffic Prediction Research Based On Support Vector Machine

Traffic prediction is one of the important means to improve the transportation management level and reduce the cost of transportation. Simultaneously, it is also the foundation of road network planning, traffic evaluation and feasibility analysis of construction projects.Therefore, the research on highway traffic prediction has important significance.Based on the analysis and comparation of various traffic prediction methods, traffic prediction method using support vector machine (SVM) and conducted practical application are studied in this paper. First, the toll data of export is preprocessed into prediction analysis data sets.Then, the grey prediction methods and neural network prediction methods are researched, and these methods are used to conduct comparison of prediction about existing data sets in the paper. Support vector machine prediction model, including data normalization, selection of kernel function, selection of model parameter etc. are deeply studied. After that, the forecasting model of traffic based on support vector machine is established. The model is used to predict the road traffic between the station of Weinan Xi and Weinan Dong in Xitong highway, the average error is limited 2.5%.Finally, the paper gave the detailed design of support vector machine forecasting method based on support vector machine.The prediction results show that the prediction of traffic using the support vector machine (SVM) is feasible and effective.The support vector machine prediction model has been applied in "comprehensive analysis and decision support system" for road resource integration project of Shaanxi province.