One dimensional Self-Organizing Maps to optimize marine patrol activities

A method for planning routes for patrol vessels is proposed. This method is based on a Self-Organizing Map (SOM) solution for the Travelling Salesman Problem (TSP), although with significant changes. The locations of reported Search and Rescue (SAR) requests, together with the locations of reported occurrences of illegal fishing activities are used as guidelines for designing the path vessel should take. However, instead of forcing the patrol routes to pass exactly in those locations, as would happen in a TSP, the proposed method uses the locations as density estimators for where the patrol effort should be placed. It then obtains a patrol route that passes through the areas with greater density. We show the behaviour of the proposed method on artificial data, and then apply the method to some data from the Portuguese Navy, obtaining possible routes for its patrol vessels.