Ship nonlinear roll motion identification using artificial neural network

The solution of the nonlinear equation for a ship's rotational motion around its longitudinal axis, even with simplifying assumptions, is complicated. This oscillatory motion, which is known as the roll motion, is gener-ated when the ship sails in the waves, and the irregular behavior of the waves causes time-varying dynamics. Calculating the ship's roll response is possible by determining roll equation coefficients. In the current study, the coefficients were determined from the dynamic response of the ship using a training feed-forward neural network. The training was carried out in two modes: as a free swing in calm water and forced oscillation in irregular waves. The DTMB 5415 vessel was selected as the case study ship. The results of the simulation by the neural network were validated by numerical analysis and model test results.