Suspected Behavior Detection Based On AIS Fishing Boats

With the increase of the number of offshore ships and the rapid development of shipping industry,maritime violations and maritime traffic accidents occur frequently.The wide application of ship automatic identification system(AIS)has brought the rapid growth of ship trajectory data.Therefore,it is possible to improve the utilization of ship trajectory data,obtain valuable information for maritime traffic and navigation,and realize the intelligent management of maritime traffic.Effective ship suspicious behavior detection algorithm is of great significance to reduce the occurrence of maritime traffic accidents and find illegal ships in time.Focusing on the research content of fishing vessel suspicious behavior detection,this paper proposes two suspicious behavior detection algorithms.One is to propose a suspicious behavior detection algorithm based on piecewise cosine similarity for the phenomenon of "one ship with multiple codes",that is,illegally applying others’ water communication service identification code(MMSI);The other is to propose an improved trajectory segment density clustering(DBSCAN)suspicious behavior detection algorithm for suspicious behavior.The main research work includes:(1)A suspicious behavior detection algorithm based on piecewise cosine similarity is proposed.Aiming at the problem of low measurement efficiency caused by the large number of sampling points of fishing vessel AIS trajectory data,an interval sampling processing method is proposed to improve the measurement efficiency;Aiming at the problems of AIS track data,such as partial track missing,longitude and latitude error and so on,a processing method of segmenting the complete track is proposed to improve the measurement accuracy;Based on the segmentation of the complete trajectory of the fishing boat,the detection accuracy is compared by setting different sampling frequencies.The experimental results show that the effective sampling interval and segmented processing can better improve the efficiency and accuracy of trajectory similarity measurement.The algorithm has certain application value in the field of trajectory similarity measurement.(2)A suspicious behavior detection algorithm for the improved trajectory segment DBSCAN is proposed.The distance calculation in the traditional DBSCAN clustering algorithm is to calculate the distance from point to point,while the distance calculation in fishing boat trajectory clustering is to calculate the distance between trajectories.This paper proposes to change the point-to-point distance in DBSCAN algorithm to segment to segment distance,At the same time,the similarity of trajectory segments is measured by combining cosine distance and improved dynamic time planning distance,which not only considers the similarity in trajectory direction,but also solves the problem of high time complexity caused by high-dimensional trajectory data.The experimental results show that the suspicious behavior detection algorithm of improved trajectory segment DBSCAN can effectively improve the quality of trajectory clustering.