In recent years,data mining technology has been widely used in the research of waterborne ship traffic,and a large amount of ship trajectory data obtained based on AIS contains ship behavioral features,how to measure the similarity between ship trajectories through these behavioral features in order to extract ship trajectories with similar movement pat-terns is the focus of the current research and the hot issue.In this paper,a new ship trajectory similarity metric is pro-posed by integrating the traditional Hausdorff distance,cosine distance and time distance from the characteristics of ship traffic flow.In order to verify the performance of the method,a comparative experiment is carried out with some ship traf-fic data of the Luotou waterway in Ningbo-Zhoushan harbor as an example.The results show that the new method pro-posed in this paper has better ship trajectory classification performance than Hausdorff distance,DTW distance,nearest-neighbor distance and LCSS distance in busy waterways,and the obtained trajectory clustering results better reflect the spatial distribution characteristics of ship traffic flow,which can be used as a reference basis for water traffic planning,traffic organization and route optimization.
NETL
NSTL
万方数据
