Research On Speed Fusion And Quality Evaluation Method Based On Multi-source Detectors

In order to improve the level of intelligent transportation system(ITS)in the context of the big data,it is necessary to have large-scale and high-quality traffic data,and make high-accuracy and reliability quality evaluation of traffic data.Multi-source detectors’ data contains rich traffic information,it will help to improve the efficiency of intelligent transportation system if used efficiently.The purpose of this thesis is to propose an efficient multi-source detector speed fusion method and a comprehensive speed quality evaluation method.Firstly,this thesis expounds the research significance of multi-source detector speed fusion and quality evaluation in the background of the intelligent transportation system(ITS),and summary the research status of data fusion and quality evaluation methods.Then,it introduces the research content and technical roadmap of this thesis.Secondly,the principle,advantages and disadvantages of microwave data,video data and floating car data acquisition technology are expouned,and the main data types that can be collected by each acquisition technology are presented.Then this thesis introduces the structure and content of the research data,and analyzes the correlation of traffic flow speed data by using Pearson correlation coefficient.Thirdly,in order to get a good result of the speed fusion,this thesis proposes a data quality control method to improve the quality of the original data;then the speed fusion model of CFNN is proposed,and four algorithms(SCG,LM,OSS,BR)are used to optimize CFNN;finally this thesis describes the speed fusion process.Then,a qualitative and quantitative method is proposed to evaluate the speed quality of multi-source detector.The qualitative evaluation method is Improved Fuzzy Comprehensive Evaluation Method based on the combined weighting method.The combined weighting method overcomes the shortcomings of the Expert Scoring Method by combining the subjective and objective weighting method.The quantitative evaluation method is TOPSIS based on triangular fuzzy number,which evaluates the speed data quality by calculating the relative distance between each data set and the ideal data set.In this thesis,the Degree of Mebership Fnction is used to combine the Improved Fuzzy Comprehensive Evaluation Method with TOPSIS to form a hybrid evaluation method to evaluate the speed of multi-source detectors,which overcomes the limitations of one single evaluation method.Finally,this thesis proves the rationality of the data quality control,fusion and evaluation method proposed by calculation examples.In the first,the abnormal data are identified and corrected;secondly,four algorithms are used to optimize the CFNN fusion model to fuse the microwave and video speed data after quality control.The results show that LM algorithm is better than SCG,OSS and Br algorithm;finally,the original data,the controled data and the fused data of microwave and video speed are evaluaed by the qualitative and quantitative quality evaluation method.The results show that the quality of the data is improved well after quality control and data fusion.