Intelligent prediction of ship hydroelastic responses in regular waves by harmonic response reconstruction principle

This paper presents a new method for the rapid estimation of ship motions and sectional loads in regular waves considering hydroelastic responses by harmonic response reconstruction principle. The present method fully utilizes the inherent physical laws behind the ship hydroelastic response of springing and whipping. A neural network-based method is utilized to learn and predict ship motions and sectional loads in regular waves from the aspect of frequency-domain accounting for the mechanical principle of ship harmonic response. The model takes the frequency, amplitude and phase of incident regular waves alongside ship forward speed as inputs, and predicts the amplitude and phase of ship motions or loads at the first several orders of harmonics, which are then used to reconstruct the time-domain responses. The presented method turns out to be useful and applicable with the examples of two ships illustration and comparison with the results by numerical calculations.