Research On Ship Anomaly Behavior Identification Based On Trajectory Clustering

Marine traffic monitoring plays an important role in the safety of underway ships.It has brought great contribution to the marine traffic administration authorities with the mandatory installation of the ship AIS and the establishment of the coastal VTS,but the main mode of marine traffic monitoring in the marine department was carried out in manual way,which is time-consuming and lack pertinence,especially in some busy port only relying on manual monitoring is far from meeting security needs.In order to real-time monitoring the ship's behavior and self-discover the ship anomaly behavior,a ship anomaly behavior identification model based on trajectory clustering is proposed.Aiming at the problem that the traditional Hausdorff distance is increasing in the case where the ship trajectories are not equal and the track points are lost,this paper optimizes the method of traditional Hausdorff distance measurement.The vertical distance in the line from track point to the nearest two points,substitute for the nearest neighbor distance,which makes distance measurement more accurate.Making Density Peak clustering algorithm applied to maritime community and clustering ship trajectories,which needn't set Parameters manually.The sweep line is set up to scan for each type of ship trajectories,and obtain a ship typical trajectory model.Calculating the deviation of distance,course and speed between each ship trajectory and ship typical trajectory,then the optimal deviation threshold is determined by the accuracy of the anomaly identification and the error rate,achieving intelligent identification of the ship anomaly behavior.Making verification through the case of ship anomaly behavior found in the Xiamen VTS monitoring center,and the experimental results show that the proposed algorithm model can identify the ship anomaly behavior effectively and find out that ship anomaly behavior can be detected earlier than manual way.It will help operators on duty provide reference for early detection of ship anomaly behavior.