ABSTRACTS OF ARTICLES OF THE JOURNAL "INFORMATION TECHNOLOGIES".
No. 11. Vol. 27. 2021

DOI: 10.17587/it.27.607-615

V. M. Grinyak, Ph.D. (Tech.), Professor, Vladivostok State University of Economics and Service, A. V. Shulenina, Senior Lecturer, Far Eastern Federal University, Vladivostok, Russian Federation

Marine Traffic Data Clustering for Ships Route Planning

This paper is about maritime safety. The system of vessel traffic schemas is one of the key elements of sea traffic control at the arias with heavy traffic. Such system based on a set of rules and guidelines defined by traffic schemas for certain water areas. From the classic approach, vessels that are not following the guidelines do not necessarily create alarming situations at the moment, however, could lead to complex danger navigation situations with the time passed. The problem of ship route planning through the area with highly intensive traffic is considered in this paper. The importance of the problem becomes more significant these days when taking in account development of self-navigating autonomous vessels. It is expected to respect area navigation limitations while planning vessel path through the areas with identified traffic schema. One of the ways to identify navigation limitations could be trajectory pattern recognition at certain sea areas based on retrospective traffic analysis.
Model representation for such task could be based on vessel moving parameters clustering. The presented model is based on solving the shortest path problem on weighted graph. There are several ways to create such weighted graphs are suggested in the paper: regular grid of vertices and edges, layer grid of vertices and edges, random grid of vertices and edges, vertices and edges identified based on retrospective data. All edges are defined as a weighted function of "desirability" of one or another vessel course for each location of sea area with consideration of identified trajectory patterns. For that the area is divided into sub areas where courses and velocity clustering is evaluated. Possible ways of clustering are discussed in the paper and the choice made in favor of subtractive clustering that does not require predefining of cluster count. Automatic Identification Systems (AIS) could be used as data source for the traffic at certain sea areas. The possibility of using AIS data available on specialized public Internet resources is shown in the paper. Although such data typically has low density, they still could well represent vessel traffic features at the certain sea area. In this paper are presenting samples of route panning for Tsugaru Straight ang Tokyo Bay.
Keywords: marine traffic control, route planning, traffic intensity, clustering, big data, graph algorithms, automatic identification system

P. 607–615

To the contents