Research And Application Of Forecasting Model Based On The Traffic Flow In Chongzun Expressway

With the development of the information technology rapidly, the industry of expressway has also brought huge economic benefits and social benefits.There are three systems which is charging system, monitoring system, communication system in the information management system of the networked toll collection,because of these systems in Chongzun expressway, it has operated8years normal, along with the construction of operation management and information system, these systems have produced a large amount of data, these data are lurking the deeper content and information beyond the surface meaning, all kinds of data which is information system of operation management are exploited effectively,which will provide the basis for operation and management and data support for operating decision strong. There are three sides:1.The theoretical knowledge of data mining was studied related,focusing on the data mining classification and prediction function, laid the foundation for the cluster analysis and ARIMA and SVM model and BP network model for data mining.2.The setting of operation management that Chongzun expressway toll collection network and some data from three systems are elaborated, the monthly traffic flow of exit and every prime-time of holiday of requirement analysis are researched, it provide data support for the study of data mining.3.Based on the traffic flow of exit and labor day during2009-2012,reaseared the forecasting the traffic flow of exit in future.Though the forecasting module of time series in SPSS was applied, firstly, judged the steady of time series using first difference; secondly, ensured the order of model using autocorrelogram and partial autocorrelogram; lastly, established the ARIMA model, to verified the feasibility of the model, the model will be applied to the actual operationto establish Teasonable toll plan, start all kinds of plans, guide operation management.4. Forecasted every prime-time traffic flow in toll station, the BP neural network in MATLAB to forecast every prime-time traffic flow in labor day was applied, the result of forecasting unsatisfactory, for this reason, SVM model was applied, and compared with the BP neural network for the prediction of traffic flow, SVM model better than BP neural network in the forecasting the traffic flow in holiday.