Discovering Urban Trip Attractive Regions Based On RFID Data Of Vehicles

Since the beginning of the new century,with the growth of the urban motor vehicle ownership and per capita possession,a series of traffic difficulties are becoming more and more prominent.Solving the urban traffic problems should be on basis of learning the traffic characteristics of the city.The urban trip attractive regions are areas with large traffic volume and frequent trips,the distribution of which is an important feature of urban traffic,enjoying more focus.The acquisition method of residents' trip information has been developing continuously,from early questionnaire survey to today's extracting information from massive taxi or bus data.However,these methods have some shortcomings.For example,there is a shortage of timeliness in questionnaire survey data,and taxi data are insufficient to reveal the overall status of urban traffic.RFID(Radio Frequency Identification)data of vehicles are the real time data collected for all motor vehicles.Good data real-time and representative of the overall traffic make the conclusion of researches using RFID data of vehicles more convincing.Chongqing is the national pilot city of RFID of vehicles,having set up a certain scale of RFID data collection points,and all local vehicles required to equip the RFID tags.Relying on this condition,this paper carried out the relevant research on the discovery of urban trip attractive regions spots by using RFID data,and put forward a set of methods.There are two main difficulties in the discovery of the urban attractive regions: 1.it is difficult to identify the stop section in the RFID trajectory;2.the spatial distribution of RFID data is discrete and sparse,not suitable for spatial clustering,which is currently the mainstream method of discovering attractive regions.This paper will focus on these two issues,including the following aspects:(1)Combining the basic theory of the trajectory and the RFID data,this paper formed the basic theory about RFID trajectory analysis,which laid a theoretical foundation for the following chapters.According to the grade division of the actual road and the layout of the RFID collection points,the grade division of the RFID road section was formed,which is convenient for discussion and classification.(2)Aiming at the problem that the stop section of RFID trajectory is hard to identify,we established the vehicle travel time distribution probability model on each RFID road section,and then combined it with Bayesian theory to determine whether there was a stop in a certain section of RFID trajectory.In this paper,the travel time distribution of the RFID road section was regarded as the joint of the travel time distribution in these three cases: ?Vehicles quickly pass through RFID road section;?Vehicles meet a traffic congestion in the RFID road section;?There is a stop in vehicles' journey.Then we used EM(Expectation Maximization)algorithm to calculate the parameters of the travel time distribution models of each RFID road section.Finally,we combined the idea of Bayes with the obtained models to determine whether there was a stop in a certain section of RFID trajectory,and compared the results with that of artificial judgment,verifying the validity of this method.(3)In view of the problem that RFID data are not suitable to spatial clustering algorithm,an attractive region detection method based on data field was proposed in this paper.In this paper,data field theory was applied to the RFID data,building the spatial-temporal distribution model of the data field in the main urban area.We regarded the areas with high data field potential as attractive regions.Then we utilized visual methods to show the urban attractive regions of different time,and combined the POI(Point of Interest)information to do some analysis.At the same time,we compared the results with that obtained by using taxi data,finding the similarities and differences between the distribution of attractive regions obtained by different data sources.