基于DBTCAN算法的船舶轨迹聚类与航路识别  

Ship trajectory clustering and route recognition based on DBTCAN algorithm

作　　者：杨家轩[1,2] 刘元 YANG Jiaxuan;LIU Yuan(Navigation College,Dalian Maritime University,Dalian 116026,Liaoning,China;Key Laboratory of Navigation Safety Guarantee of Liaoning Province,Dalian Maritime University,Dalian 116026,Liaoning,China)

机构地区：[1]大连海事大学航海学院,辽宁大连116026 [2]大连海事大学辽宁省航海安全保障重点实验室,辽宁大连116026

出　　处：《上海海事大学学报》2022年第3期7-12,共6页Journal of Shanghai Maritime University

基　　金：国家自然科学基金(41861144014,41501490)。

摘　　要：为解决船舶轨迹聚类算法效率不高,检测精度低,丢失轨迹局部特征等问题,将具有噪声的基于密度的空间聚类(density-based spatial clustering of applications with noise,DBSCAN)算法由传统的点聚类推广为线聚类,提出一种可以直接对完整船舶轨迹进行聚类的具有噪声的基于密度的轨迹聚类(density-based trajectory clustering of applications with noise,DBTCAN)算法。该算法采用Hausdorff距离作为船舶轨迹之间的相似度度量,可以对不同长度的船舶轨迹进行聚类。针对DBTCAN算法需要人工确定输入参数的问题,提出一种参数自适应确定方法。选取渤海海域的船舶自动识别系统(automatic identification system,AIS)数据进行实验,结果表明,该算法能够在大量复杂的船舶轨迹中找到相似的轨迹并对其进行聚类,聚类结果与实际交通流情况一致。本文的研究成果可以为相关部门进行航线规划和海上交通监管提供依据。In order to solve the problems of low efficiency,low detection accuracy,and loss of local trajectory features of ship trajectory clustering algorithms,the DBSCAN(density-based spatial clustering of applications with noise)algorithm is extended from point clustering to line clustering,and a DBTCAN(density-based trajectory clustering of applications with noise)algorithm that can directly cluster the complete ship trajectory is proposed.In the algorithm,the Hausdorff distance is used to measure the similarity of ship trajectories,and the clustering of ship trajectories with different lengths can be achieved by the algorithm.An adaptive parameter determination method is proposed for the problem that DBTCAN algorithm requires manual determination of input parameters.Taking the AIS(automatic identification system)data of the Bohai Sea waters for experiments,the results show that DBTCAN algorithm can find similar trajectories and group similar trajectories together into a cluster from a large number of complex ship trajectories,and the clustering results are consistent with the actual maritime traffic flow.The research results can provide a basis for relevant departments to carry out route planning and maritime traffic supervision.

关 键 词：船舶轨迹聚类 具有噪声的基于密度的轨迹聚类(DBTCAN) HAUSDORFF距离 自适应参数 航路识别 

分 类 号：U697.3[交通运输工程—港口、海岸及近海工程] U675.79[交通运输工程—船舶与海洋工程]