Research On Vehicle Trajectory Abnormal Detection Method

In recent years,vehicle trajectory anomaly detection is one of the important indicators reflecting the traffic conditions of expressways.It is also the traffic information that travelers pay more attention to.If we can predict the future abnormal behavior of the vehicle in advance,it will not only allow the highway traveler to select a reasonable driving plan more quickly,but also avoid the congestion of road sections and dangers and reduce fuel and time loss.It can also provide managers with reasonable decision-making basis for traffic control and management.In this paper,we study and discuss the problem of vehicle trajectory anomaly detection combined with trajectory clustering analysis.The main innovations are summarized as follows:(1)Based on the spectral clustering process of multi-feature parameter modification,combined with spectral clustering and modified parameters,a feasible scheme for vehicle trajectory spectrum clustering is given.Firstly,the application problems of trajectory similarity measurement methods under different vehicle motion speeds are analyzed and discussed.Secondly,combined with the vehicle abnormal behavior recognition flow chart,the anomaly detection and recognition of the vehicle trajectory coordinate position in the two-dimensional space is completed on the premise of completing the spectral clustering.Finally,the specific experimental methods are given for the proposed algorithm,and the effectiveness of the method is verified.(2)Aiming at the problem of abnormal trajectory detection in vehicle trajectory,a vehicle abnormal trajectory detection algorithm based on DBSCAN algorithm is proposed.First,the trajectory preprocessing process and the density clustering algorithm are described.Secondly,the method of similarity measurement,Hausdorff distance calculation method,is described.Then the trajectory clustering analysis process is analyzed and described,and the vehicle abnormal trajectory detection algorithm based on DBSCAN algorithm is described in detail.Finally,on the actual trajectory set,the experimental verification and analysis of the trajectory clustering accuracy and the abnormal trajectory detection are carried out,and the experimental results are analyzed and explained.