Interval Prediction Of Origin-Destination Matrix Based On Expressway Network Toll Data

Expressway OD matrix prediction is one of the links of expressway traffic demand prediction and an important basis for expressway network planning.However,the acquisition of base year OD matrix is a difficult problem in the OD matrix prediction process.With the development of traffic big data,The expressway network toll data provides a new opportunity for the base year OD matrix acquisition and vehicle spatio-temporal characteristics analysis.Owing to the numerous uncertainties in traffic travel,higher requirements are imposed on OD matrix prediction.In order to meet the requirements of highway network planning and management,it is imperative to study the OD matrix prediction of expressway vehicle travel.At the same time,in order to reduce the risk of decision-making,it is necessary to conduct interval OD matrix prediction research.Based on the background of traffic data,this thesis conducts pre-processing work such as data cleaning,data integration and toll station numbering for data collected by networked toll.By extracting the current OD matrix,the time-space characteristics and vehicle structure analysis of expressway vehicles are analyzed.On this basis,a BP neural network interval combination forecasting method based on the network minimum error is proposed.Then,aiming at the minimum difference between the actual value and the predicted value,the difference between the midpoint and the radius is used to construct the traffic occurrence and the combined interval prediction model.In order to clarify the trend of the target OD travel volume,the concept of trend coefficient is proposed,and the key of OD matrix prediction is transformed into the solution of interval trend coefficient matrix.Taking the current OD matrix point multiplication interval trend coefficient matrix to get the row sum,column sum of the new matrix and the shortest interval distance of the target annual traffic volume and attraction as the objective,the interval OD matrix prediction model is established.Finally,based on the data of expressway network toll collection of G province,the OD matrix of the expressway of this province is predicted by the method of this thesis,the rationality of matrix distribution is tested according to the results of current data analysis.The interval OD matrix prediction method based on expressway network toll data proposed in this thesis can fully consider the uncertainty of traffic travel.The predicted interval OD matrix can provide theoretical support for expressway network planning and operation management.