Redes neurais artificias aplicadas à análise do comportamento de sistemas oceânicos

Abstract

The present work develops research involving the application of artificial neural networks (ANNs) in the prediction of the ship's roll movement behavior, operating in an irregular sea, focus to the amplification and decrease of roll of vessels due to the phenomenon of parametric resonance. The importance of this topic, parametric resonance excited by longitudinal waves, is a real problem that affects many ships and can lead to catastrophic consequences. For this, a code was developed in MATLAB based on RNAs capable of making fast predictions, and sufficiently reliable of the progression of exaggerated movements. Previously, the prediction of the behavior of the real sea itself, an element that induces the movement of oceanic systems, was analyzed. As an illustration, it is discussed the application in the prediction of the vertical displacement and the excitation force of a WEC (Wave Energy Converter) system. The objectives of the thesis could be demonstrated through the analysis of the predictions of the behavior of a container ship, operating in the real sea, covering windows of long prediction time, sufficient (~ 60s) to enable corrective measures to prevent the appearance of the parametric roll. The practical motivation of the thesis intended to provide elements for the installation on board an intelligent online system, to guide the crew in detecting the appearance of parametric resonance, preventing this dangerous phenomenon from being able to develop.