Research On Traffic Status Evaluation Based On Traffic Data Fusio

Real-time and accurate road traffic information proves to be an important guarantee for modern traffic management and control,which is also an important element of construction of intelligent traffic system.Evaluating road traffic flow data objectively and quantitatively has been one of the hottest research topics.However,single collection technology of traffic data are vulnerable to some own factors and have some limitations in collection method,data parameters,sample size,collection range and data accuracy.Therefore,evaluation of road traffic by data acquired by single detection technology cannot correctly reflect traffic state of a certain time and significant errors have been resulted.On the other hand,fusion of Multi-source data traffic by data fusion technology could make up some disadvantages of single data source,achieving mutual complementarity and acquiring more precise traffic data,thus improving accuracy of the estimation.In this paper,fusion problems of two data sources were investigated based on characteristics of collection data of floating car and fixed detector technology.A scheme of real-time estimation of road traffic state was proposed and data fusion model based on index of road traffic state was established by using improved BP NN algorithm.First,technology of traffic data detection and data fusion was studied in detail.Corresponding treating methods were proposed in term of the problems found in fixed detector data and floating car data,including mending methods based on time series analysis and Kalman Filtering repairing Model improve by data fusion.Also,match of time and space of data was discussed.Then,based on research of traffic index at home and abroad,computation model of traffic state index was established and data fusion model based on index of road traffic state by BPNN improved by momentum method.Finally,simulation was carried out based on roads of Wuhua district in Kunming and obtained traffic data was calculated and analyzed,making it possible to evaluate traffic state index.Results show that there is a dramatic increase of traffic state index obtained by fusion in terms of data precision and stability,validating reliability and effectiveness of the model.