The Application Research Of Four Stages Forecast Method Based On Neural Network In Traffic Volume Prediction

As the high-speed development of economy, lots of phenomenal achievements of transportation and road construction have been made. With the strong support of national policies, the construction of transportation develops fast, the level of road technology grade is constantly improved and the road network is being perfected step by step. Among them, the analysis and prediction of traffic volume played essential function in the process.The analysis and prediction of traffic volume is a scientific method which speculates the traffic volume for the allocated routes in the future. This method is comprehensive considering the past and present changes and the development of the factors which have relation with traffic, economy and society in some area. This forecast is the basis for confirming the technical standard of highway construction, engineering facilities scale and economic evaluation. It should consider the social economic status, land utilization, comprehensive transportation situation of effecting area and their development planning, predict the future traffic which aiming project may bear of in the certain period according to the principle of combining quantitative calculation and qualitative analysis, and provide the basis for selecting highway technical standards, comparing route scheme, analyzing economic financial benefit and other works.At present, there are many traffic volume prediction methods, and the related theory become mature. But the practices show that all of the predicting methods have their own disadvantages. Through the studying, analysis and comparison of traffic volume prediction, this essay proposed that using the method combining the improved four stages forecast method and neural network to predict traffic volume, so as to get the purposes of improving prediction accuracy, fully considering fuzzy parameter related prediction, and closing to actual engineering application.This forecast model adopted four stages forecast method as the main body, respectively done some analysis of social economic forecasting, occurrence attract traffic volume forecasting, traffic distribution forecasting, transportation share forecasting, traffic assignment forecasting, lure traffic flow and transfer traffic, selected suitable prediction model or method to predict. Among them, using the quantitative and qualitative analysis for social economic forecasting, using neural network for occurrence attract traffic volume forecasting, using Frator Method for traffic distribution forecasting, and transportation share forecasting should considering multiple transportation way, using the capacity limit increment of load multipart probability assignment method for traffic assignment forecasting, and the lure traffic flow and transfer traffic uses unconstrained gravity model.This essay applies this forecast model to traffic volume prediction of Enshi outer ring, through the investigation and analysis of local actual condition, considering all effecting factors, full using the characteristics of neural network nonlinear processing and fault tolerance power, analyzes the effecting to the aiming project of transportation way in the region and gets the prediction results. Through the comparing of the prediction results and the development situation of actual economy, society and traffic in local place, the results conform the local development trend and have certain reference value.