Study On Data Fusion And State Identification Technology Of Urban Traffic

At present,with the rapid development of science and technology,such as intelligent transportation system,internet+,cloud computing of big data and road-vehicle cooperation network,the effective collection,processing and integration of traffic data,timely and accurate prediction and identification of traffic conditions are vital for traffic management departments to implement real-time effective traffic control measures.The paper is aimed to compare and analyze the collection,processing and fusion technology of traffic data,traffic state-identification technology and related applications internal and abroad,and to discuss the improvement of related technology in the light of the specific situation of road traffic in China.The paper is expected for the following effects: first,to enrich technical theory,main part of which is about to improve the relevant algorithm model to meet applying needs in different conditions.Second,to improve the practical applying effects by correcting possible errors in the application of the technology and technical supplement in order to provide more accurate real-time traffic data for the traffic management department.Firstly,the three key technologies as traffic data collection,preprocessing and data matching are studied in detail.A comprehensive analysis sheet on technical advantages and applicability of different data collection methods is made.Algorithms and methods of data preprocessing and data matching have been improved and a framework of pre-integration data processing and an optimized technical system has been built up.Secondly,a variety of internal and abroad data fusion algorithms and models are studied in depth,and a sheet of advantages and disadvantages as well as applying characteristics of various algorithm models is shown.Based on BP neural network,an improved fusion model and its sub-model are proposed according to practical applications,all parts of which the network structure design and parameter setting are completed at the same time.Then,based on a large number of basic traffic flow data,from microwave detector,coil detector,video detector,floating vehicle and self-test license detection,and collected by Zhengzhou multi-source traffic detector,the improved algorithm model has been tested through instances.The results obtained before and after the fusion and the different fusion sub-models are quantitatively compared,and are proved its validity and accuracy by comparing the real data.Finally,the various traffic-state-identification algorithms and models are studied and analyzed.Based on the analysis of the traffic data of Longhai expressway in Zhengzhou,different road conditions are classified,the specific limit parameters are determined,and the simulation model of the relevant road is established by using Vissim software.The threshold algorithm of abnormal state identification is obtained by simulation experiment under different conditions of road section.