Research On The Travel Behavior Of Private Car Users In Urban Enviroment

In the era of big data,it is of great significance for the government and traffic management departments to formulate scientific and reasonable urban management strategies by using vehicle trajectory data and carrying out related research on residents’ travel behavior.But most of the research work is focused on the field of public transportation,and the research on the data of private cars is relatively few.In order to make up for the missing,it is necessary to study the travel behavior of private car user groups in cities based on the private car trajectory data.Based on the real data set of private car,this paper analyzes and studies the travel behavior of private car users.In order to obtain the basic data from the original private car data set to satisfy this research,the private car data set preprocessing framework is proposed to preprocess the collected original private car data set.First,a method is proposed to replace all redundant points with a continuous mean of several redundant points to clean the redundant data in the GPS trajectory data;then,for the noise data existing in the GPS trajectory,the noise data is eliminated by the method called heuristics-based outlier detection;finally,based on the corresponding time information in the GPS trajectory data and vehicle driving state data,two kinds of data are fused to obtain the private car trip and stop trajectory data.In order to grasp the travel behavior laws of private car users,an analysis method based on the basic attributes statistics of private car users is proposed to study the travel behavior of private car users.First of all,through the statistics and analysis of statistical results of basic attributes which describing travel move behavior of private car users,such as move distance,obtaining the overall move laws of private car users;secondly,by statistics and analysis of statistical results of basic attributes which describing the private car users travel stop behaviors,such as stop duration,obtaining the overall stop laws of household group.At the same time,by fitting the probability density of move distance,move duration and stop duration,it is found that these important attributes obey different probability density distributions in working days and weekends,respectively.In order to determine whether private car users are regular travelers,a semi-supervised algorithm(TFC2-means)is proposed.The algorithm is divided into classification module and decision module.In classification module of the algorithm,the travel behavior characteristics of private car users are first calculated from the trip and stop trajectory data of individual users according to the feature extraction formula.Then,the supervised information is obtained from the labeled samples,which is to add constraints to the training process,and the private car user groups are divided according to the travel behavior characteristics and constraint information.In determination module of the algorithm,according to the group division results determine whether private car users are regular travelers.The Experiment is based on the real private car travel data in the city.Compared the TFC2-means with the TAD algorithm,the accuracy,precision and recall were increased by 33%,19.07% and 48%,respectively,to verify the rationality of TFC2-means algorithm selection features.In addtion,the TFC2-means was compared with the traditional K-means method,and the accuracy,precision and recall were increased by12.5%,8.12% and 22%,respectively,prove the effectiveness of TFC2-means algorithm constraint addition.