Using Dominant-Set clustering to characterize vessel pathway in port areas

Abstract

Vessel pattern identification helps traffic management and collision avoidance in port areas. Suggested routes for ships, also called certified paths, can be extracted by performing data mining techniques on Automatic Identification System (AIS) data. The certified paths are fundamental to properly monitor vessels during their journeys in order to study marine traffic in port areas. Extracting these routes represents the main subject of this thesis. This type of analysis is traditionally performed through clustering, providing a method to automatically extract certified paths given a collections of trajectory data. In this project, the Dominant Set clustering algorithm is applied on vessel trajectories to retrieve these routes. After the process of certified path extraction, a set of points (waypoints) describing them is automatically found. These points are then provided to the vessels approaching or leaving the seaport. Each of them contains information about the allowable areas for the vessels and the time intervals in which ships must approach them to ensure a smooth and safe behaviour in port area. The results obtained shown the superiority of the DS over well known alternatives.