船舶轨迹快速相似度度量与改进自适应密度聚类

Fast Similarity Measurement of Ship Trajectory and Improved Adaptive Density Clustering

船舶自动识别系统(AIS)数据蕴藏着大量海上交通特征.为挖掘出AIS交通流数据中有关船舶运动规律有效的、潜在的特征信息,文中提出一种新型船舶轨迹间相似度度量方法和基于核密度估计的改进自适应密度聚类以提取交汇水域航道内船舶交通流轨迹特征.将船号MMSI作为信息节点生成船舶轨迹,采用文中快速相似度度量算法对轨迹进行相似度度量,应用改进自适应密度聚类算法结合快速相似度度量算法对船舶轨迹进行聚类得出不同船舶轨迹类簇.以长江口南港南槽航道交汇处水域为例,采用文中新型快速相似度度量算法极大的缩短了船舶轨迹间相似度量时间,采用改进自适应密度聚类算法弥补了密度聚类中参数难以确定的问题,提高了聚类效果缩短了聚类时间.

Automatic identification system(AIS)data contains a lot of maritime traffic characteristics.In order to dig out the effective and potential characteristic information about the ship motion law in AIS traffic flow data,this paper proposed a new similarity measurement method between ship trajectories and an improved adaptive density clustering based on kernel density estimation to extract the characteristics of ship traffic flow trajectories in the channel of the intersection waters.Taking the ship number MMSI as the information node to generate the ship trajectory,the fast similarity measurement algorithm in this paper was adopted to measure the similarity of the trajectory,and the improved adaptive density clustering algorithm combined with the fast similarity measurement algorithm was used to cluster the ship trajectory to obtain different ship trajectory clusters.Taking the waters at the intersection of South Channel and South Channel of the Yangtze Estuary as an example,the new fast similarity measurement algorithm in this paper has greatly shortened the similarity measurement time between ship tracks.The improved adaptive density clustering algorithm makes up for the difficulty in determining the parameters in density clustering,improves the clustering effect and shortens the clustering time.