Research On Traffic State Identification Of Urban Road Section Unit Based On Different Tytes Of Floating Car Data Fusion

Accurate identification of the traffic state of road units is the key to formulate a dynamic traffic control scheme and alleviate traffic congestion.Traffic congestion in urban roads usually begins at a local bottleneck section of the road.As the gap between traffic supply and demand increases gradually,traffic congestion intensifies,and congestion gradually spreads to the surrounding roads.It is very important to recognize the traffic congestion of the bottleneck section in time and accurately for improving the quality of road traffic operation.GPS floating car data acquisition technology as a new and rapid development of data mobile acquisition method,compared with the traditional fixed detector,can collect road traffic information in an all-round way.Influenced by the complex environment of urban roads,the running state of vehicles in the road varies.The traditional data of single-vehicle floating cars only represent one kind of running vehicles in the road.It is one-sided to describe the running state of the whole section with the running state of a single type of vehicle,which is not convincing enough.Compared with the single floating car data,different types of floating car data reflect more types of vehicle operation information in the road,which is more comprehensive in describing the traffic flow operation state.Therefore,it is necessary to fuse the different types of floating car data,and to study the method of traffic condition identification based on different types of floating car data fusion.Taking three different types of floating car data which includs taxi floating car,bus floating car and private car floating car data as data sources,this paper uses FCM algorithm to determine the traffic state partition method under floating car data.In view of the fact that a single floating car data source can only reflect the traffic state of one type of vehicle on the road,and can not fully reflect the traffic state of the road,and the inconsistency of traffic state classification standards for different types of vehicle data on the road,the D-S evidence theory algorithm in data fusion technology is introduced to fuse the different types of floating car data,and constructs a method to distinguish the traffic state of road segment unit based on D-S evidence theory.Aiming at the shortcomings of D-S evidence theory in dealing with highly conflicting data,a new D-S evidence theory improvement method is proposed,which combines the modified basic trust allocation function of evidence source with the optimized synthesis rule.By studying the relationship between the sample size of floating car and the reliability of traffic state identification results,an improved model of traffic state identification for section units is constructed based on the improved D-S evidence theory.The main contents of this paper are as follows:(1)Research on traffic state partition method based on floating car data.The running characteristics of different types of vehicles in road section unit are analyzed from two aspects: vehicle running speed and speed dispersion.The results show that there are obvious differences in the running characteristics of different types of vehicles even in the same traffic state.Combining the fuzziness of traffic state and the difference of operation characteristics of different types of vehicles,a traffic state partition method based on FCM algorithm for different types of floating car data sources is proposed.(2)Research on traffic state identification method of road section unit based on different types of floating car data fusion.In order to solve the inconsistency of traffic state classification criteria for different types of vehicle data,effectively fuse the different types of floating car data on the section unit,and achieve a comprehensive description of the traffic state of section unit,the D-S evidence theory in data fusion is introduced to construct traffic state identification model of the section unit based on the D-S evidence theory.(3)The relationship between the sample size of different types of floating car and the reliability of traffic state identification results is studied.The scatter relation between the sample size of different types of floating car and the reliability of traffic state identification results is obtained by random sampling method.In SPSS software,logarithmic function,inverse function and power function are used to fit and compare the scatter relation.The results show that the correlation coefficient of logarithmic function fitting is the highest,and the relationship between sample size and reliability of traffic state identification results is logarithmic function.When the reliability of traffic state identification results is 100%,the minimum sample size of taxi,bus and private car data used for traffic state identification of section unit is 18,16 and 25 respectively.At the same time,when the sample size is less than the minimum sample size,the specific functional relationship between the sample size of three types of floating car and the reliability of traffic state identification results is obtained.(4)An improved model for identifying the traffic state of road units is constructed.Aiming at the problem that D-S evidence theory can not accurately fuse highly conflicting multi-source data,a new D-S evidence theory improvement method is proposed,which combines modified evidence source basic trust allocation function and optimized synthesis rules.Based on the improved D-S evidence theory,using the reliability of traffic state identification results of floating cars as the correction factor of the basic trust allocation function,an improved model of traffic state identification of section units is constructed.The improved model is analyzed by typical cases.The improved model is analyzed by typical cases,and the results show that the improved model can effectively solve the problems of traditional D-S evidence theory in traffic state identification with high conflict data sources.(5)Example verification.Taking a section unit of Chongqing Expressway as an example,this paper compares the improved traffic state identification model of road section unit,the identification model based on D-S evidence theory and the identification model based on a single data source(taxi,bus,private car).In terms of identification effect,the identification accuracy based on the improved model is 94.03%,the identification accuracy based on D-S evidence theory is 85.07%,and the identification accuracy based on taxi,bus and private car data sources is 83.58%,70.15% and 61.19%,respectively.It is proved that the application of D-S evidence theory in traffic state identification of road section unit can effectively fuse different types of floating car data information and improve the accuracy of the identification results.Compared with the model based on D-S evidence theory,the improved model can further improve the accuracy of the identification results.