Ship weather routing featuring w-MOEA/D and uncertainty handling

Abstract

The paper presents a new version of evolutionary multi-objective weather routing (WR) for ships taking into account uncertainties of weather forecasts in route optimization. The method applies authors’ w-MOEA/D algorithm: MOEA/D framework incorporating Decision Maker’s (DM) preferences by means of w-dominance relation. Owing to this, DM preferences are taken into account throughout optimization, allowing the process to focus on the part of vast objective’s space. Only the part of Pareto front being of interest to DM is generated, thus the process converges faster, without sacrificing quality of the final set. All of the above is essential for the WR method, which pursues three objectives while trying to meet multiple constraints and handling uncertainty of weather data. The final method has been implemented as a part of client–server system architecture, whose client part has been installed on board of a m/v Monte da Guia (MdG) vessel navigating between the Portuguese coast and the Azores. The method has then been verified in the course of computer simulations and its results have been compared with real MdG GPS routes. The comparison shows that the presented method is able to find routes that bring progress in terms of the objectives’ while satisfying the constraints.