Application of Machine Learning for Prediction of Wave-induced Ship Motion

Abstract

The main objective is to investigate the possibility that machine learning can be used in the real-time simulation of ship motion. A short-term prediction of seakeeping and maneuvering is strongly required for the navigation process. However, accurate and instant prediction remains still challenging. In the present study, a physics-based machine learning model has been proposed. Two physical attributes that define the ship motion have been learned by applying a machine learning model suitable for each character and integrating them. The integrated machine learning model has successfully learned the motion characteristics of the vessel.