Mining Ship Trajectory Information Based On AIS Data And Its Application

With the development of the reform and opening up,especially the proposal of “One Belt and One Road” policy,the volume and frequency of Chinese maritime transport is increasingly heavy,and the maritime traffic situation is more and more complex.In order to guarantee the safety of navigation,it was very important to strengthen the understanding and supervision of maritime traffic patterns in specific area.For a long time,the method to obtain route information was mainly by electronic chart with fuzzy qualitative analysis,lacking of specific attribute and practical navigation experience of ship.The establishment of AIS(Automatic Identification System)and IMO regulations on it could provide plenty of navigation information,which provided new ideas for quantitative analysis of maritime route.Currently,AIS data include ship name,identify number,position,speed,direction and other information,and the update cycle is short,which makes it possible for the construction of ship’s trajectory.This paper used the AIS data to construct the ship’s trajectory and adopted the clustering algorithm based on the attributive similarity to mine the ship’s trajectory,and the main work is as follows:(1)Preprocess AIS data.Considering the problems of AIS data in reconstructing ship’s trajectory,this paper adopted position,time and quantity information to clean data,historical data for interpolation,and RDP algorithm for data compression.While maintaining the original characteristics,redundant data were simplified to improve the efficiency of data analysis;(2)Optimize the similarity measurement between trajectories.Considered the characteristics of ship’s trajectories,Euclidean distance and Hausdorff distance were used to establish similarity measurement between ships’ position,direction and speed,and weighted sum of attributive similarity.The attributes of ship’s trajectories on clustering results were comprehensively considered;(3)Establish a clustering method based on attributive similarity measurement.Based on DBSCAN algorithm,a clustering algorithm with structural similarity of trajectories was formulated.In addition,the parameter selection of original algorithm was changed,membership function and local anomaly factor were added,and the clustering results were used to construct typical trajectories of ships and detect abnormal behaviors;(4)The AIS data of cargo ships in Shanghai port were used for experiment.The results showed that the proposed clustering algorithm could effectively perform clustering analysis on complex and redundant AIS data in specific segment,and the clustering results consisted with the reality.The clustering results could help in route design by navigation experience,anomaly detection and other scenarios.The proposal method and clustering results could provide support for the implement of “One Belt and One Road” policy in maritime supervision.