一种基于改进的DBSCAN的面向海量船舶位置数据码头挖掘算法

目前我国正在大力推行"一带一路"航海战略,航海事业蓬勃发展,大量新码头正在修建中。如何快速、准确地更新码头的空间信息,对于分析进出口贸易、提高码头服务效率等具有很强的现实意义。当前我国主要通过人工测绘手段更新海图,更新间隔在3~12月,远不能满足需求。而利用包括国际海事卫星C系统、北斗卫星、Argos卫星等手段获取的船舶位置数据来进行码头挖掘,为解决获得码头空间信息问题提供了新手段。利用自动识别系统AIS获取的海量船舶位置数据,提出了一种基于自优化参数的码头挖掘算法DBSCAN。一方面能够面向不同船舶类型的不同密度分布进行自动学习优化DBSCAN核心参数,进而聚类出包含码头的停泊区域,具备很强的灵活性;另一方面,融合岸基结构物等空间数据,对停泊区域中的锚区和临时停泊区域等进行排除,获取码头的空间信息,并且达到很高的准确率。利用2012年4月至2014年4月两年中国滚装船的真实轨迹数据和国际滚装船真实轨迹数据进行了码头挖掘实验,准确率能够达到93%以上。

A dock mining algorithm for massive vessel location data based on improved DBSCAN

In the background of "One belt, One road " maritime strategy in current China, maritime industry is booming, and a lot of new docks are under construction. How to quickly and accurately update the space data of wharfs has a strong practical significance for analyzing foreign trades and improving the efficiency of port services. Artificial means are currently the major approach to update marine charts, but an update interval is usually between 3 and 12 months, which cannot meet the demand. As a new approach to obtain space information of docks, we can apply Including Inmarsat C system, Beidou satellite, Argos satellite and other means to obtain vessel position data for dock mining. We propose a dock mining algorithm based on self optimizing parameters DBSCAN (density based clustering) for large scale vessel position data from Automatic Identification System (AIS ). On the one hand, the DBSCAN core parameters of different vessel types with different density distributions can be automatically optimized and learnt, and the mooring areas containing docks then can be mined out flexibly. On the other hand, by fusing with the integration of land based structures and other spatial data, we can exclude anchor zones in mooring regions, berthing areas and etc, with high accuracy. Experiments on Chinese real locus of Ro Ro real data and the international trajectory data over the past two years (April 2012~April 2014) show that the proposed algorithm can mine docks with an accuracy rate of 93%.