Research On Repair Methods For Urban Expressway Traffic Flow Data

ITS (Intelligent traffic system) is a total definition of new road traffic system involved drivers and governors, which includes vehicle, road and circumstance. Traffic information processing by various traffic analysis models is the core problem in ITS. Traffic analysis model is built on the basis of large volume of traffic flow data. The purpose of this paper is to improve the quality of traffic flow parameters by data repair, and to ensure the accuracy of traffic analysis model and the validity of ITS.This paper mainly aims at the problem that the accuracy of original data from the detection system of urban expressway is insufficient. After depth analysis of traffic flow characteristics and objective conditions to cause data defect, this paper builds up the relational model among the three traffic flow parameters, and then screens abnormal data. At last, the paper proposes two repair algorithms for expressway data defects.The traffic flow data repair algorithm based on the statistical correlation analysis is established on the improvement of the time and the spatial correlation data repair methods which are most commonly used. Through the precise numerical calculus, the definite relevant biggest reference data set carries on the repair to the damage data. Through precise numerical calculation to determine the most appropriate data of the most relevant to do date modification.The traffic flow data repair algorithm based on BP Neural Network builds up a 3D surface model about three traffic flow parameters, and then realizes the traffic data repair according to the principle of surface repair through Back-Propagation Neural Network.Carrying on the experiment through the actual sampled data, and carrying on the analysis and the comparison to the result, these two traffic flow data repair methods can both achieve satisfaction. The difference lies in the object type to repair. Choose the appropriate data repair algorithm according to the different situation, can hand over the data repair problems by the easiest and fastest method, and can achieves accuracy requirement to meet the need of traffic management systems.