Parametric Identification of Nonlinear Ship Roll Motion from Forced Roll Data

Abstract

A method of estimating the parameters in a nonlinear, single-degree-of-freedom model of ship roll motion is described. It is shown that the "linear-in-the-parameters" nature of the model allows formulation of a relatively simple, and computationally very efficient, estimation procedure, based on a recursive, linear least-squares algorithm. The method is applied to the case of forced roll, where the roll moment excitation can be accurately determined. The general nature of the estimation scheme allows the combining of data from a sequence of experiments with sinusoidal forcing of varying frequency, and also allows the use of data arising from a nonsinusoidal forcing moment. In the latter case all the required parameters in the ship motion model can be estimated from a single test. The technique is illustrated by applying it to both digitally simulated data (where the parameters are known, a priori) and to real data, obtained from a model ship. In the latter case the parameter values obtained by the proposed method are shown to agree well with estimates obtained by alternative, independent methods.