AIS Data Oriented Ships' Trajectory Mining and Forecasting Based on Trajectory Delimiter
Ya-lun Zhang, Pengfei Peng, Jian-shu Liu, Shu-kan Liu · 2018
Studying and analyzing the navigation trajectory pattern of a specific target ship or target cluster can provide important help for maritime traffic management, channel clustering analysis, and monitoring of key sea area. Therefore, this research has gradually become a necessary practical need. In order to analyze the ship's trajectory and master the rule of ship's navigation and action, this paper firstly introduces the trajectory delimiter based on the three dimensions of time, longitude and latitude to divide the ship's trajectory into different phases of navigation, and generate a trajectory transactional databases that can be used for trajectory pattern mining. Then, the frequent sequential pattern mining technique is used to extract the frequently used trajectory of marine vessels and the association rules that can be used for trajectory prediction are obtained. Using the real AIS data testing algorithms, the experimental results show that this method can effectively excavate the typical trajectory sequence of the ship and avoid the impact of sudden situations such as interference points and temporary change of channel. The association rules generated from the frequent trajectory sequence patterns can also be used for the trajectory forecasting.