Time Series Linearization of Nonlinear Ship Structural Response using the NARX Model

In this paper, a method to linearize the nonlinear response of Ultra Large Container Ship(ULCS) with a bow flare is studied by using the Nonlinear Auto-Regressive with External inputs(NARX) model. Response linearization is closely related with the development of digital twin model of a ship, where ocean wave spectrum around sea-going vessel is trageted based upon linear theory between wave and ship response. Linearization scheme is based on the system identification approach using NARX(Nonlinear Autoregressive with External Input) method, which is popularly used for nonlinear dynamic system identification. For the acquisition of data, the nonlinear time domain analysis was performed, where the nonlinear Froude-Krylov force and restoring force and restoring force of the external forces acting on the vessel were considered. Obtained structural response, such as vertical bending moment at the midship section of the ship, was analyzed using NARX, and linear transfer function was extracted using harmonic probing method. To validate the accuracy of the proposed methodology, a structural response in irregular waves was computed in time domain and compared with the linearized responses obtained from transfer function and probability of peak sagging moment exceedance. In terms of the probability exceedance peak responses, the linearized responses turned out to be in good accordance with direct linear analysis results.