Machine learning applied to ship maneuvering simulations.

• Autor(a) principal: Moreno, Felipe Marino
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/3/3152/tde-18052021-142324/
• topic: Aprendizado computacional; Cluster analysis; Hidrovias; Machine learning; Maritime simulation; Operação naval

With the increase of computational power, ship maneuvering simulations have become an important tool to improve the safety of the operations carried at the sea. In this context, one of the most important categories of simulations made by the Numerical Offshore Tank (TPN-USP) is the Real-Time simulations, carried out in a Virtual Reality environment at the same time scale as a real maneuver. These simulations are used to evaluate maritime maneuvers\' risks and viability, but since they take a long time, only a few can be made per day. This work focuses on applying machine learning to create a tool for the TPN-USP maritime simulator that will be used to choose environmental conditions of wind, currents, local sea waves and swell for these simulations.