A Method Of Clone Vehicle Identification Based On Data Mining And Time-space Characteristic Analysis Of Vehicle Trajectory

With growing number of motor vehicles,it has become a problem that vehicle-related crimes is increasing,which causes all kinds of social problem.Identifying deck vehicles with efficiency and accuracy has become an urgent task for traffic law enforcement agencies.In view of the fact that currently there are no scientifically intelligent deck car analysis using the electronic cameras,smart bayonet and electronic police devices on the urban roads,and the existing detection methods for deck cars have high costs and low detection efficiency.Disadvantages.This paper explores the practical application of data mining technology in the "deck" phenomenon,and has studied the identification methods of three deck suspect vehicles,which are:1.Based on travel time and space contradictions deck vehicle identification method.2.The deck vehicle identification method based on the contradictory path planning of Gaode map and the actual driving behavior.3.A method of clone vehicle identification based on data mining and time-space characteristic analysis of vehicle trajectory For the problem of deck vehicle identification in urban road traffic research,a large number of passing records collected by the existing traffic bayonet cameras are used as experimental data.First,the car records are extracted and the related data cleaning work is completed.Secondly,picking up travel time from two adjacent passing records is short,travel distances are far apart,and time-space conflicting passing data appears.Then,according to the driving path planning function,the real passing trajectory of the vehicle is simulated,and the number of “leakage” points existing in the simulated real driving route is recorded.Furthermore,they are used as vehicle space-time feature data,and data mining techniques are used to use the support vector machine algorithm to train the deck car classifier model.Finally,all the data to be measured are input into the trained classifier model to classify and determine the deck suspect vehicles.The experiment uses the SVM algorithm to train the identification model of the deck car,and carries out five-fold cross validation on the sample data to eliminate the problem of classifier over-fitting.The inspection method of the deck car combining data mining technology and vehicle space-time feature analysis makes up for the defects of the traditional manual inspection deck car.This model can better screen suspected deck vehicles and provide a reliable reference for the traffic control department.The work intensity of the inspectors has been reduced,the investment in testing costs has been avoided,the analysis has been more scientific and efficient,and the scope of inspections has been extensive and the results have been better.