PARAMETRIC IDENTIFICATION OF MANOEUVRING MODELS FOR SHIPS

Time domain simulation techniques are becoming a widely accepted tool for the evaluation of controllability and manoeuvring performance of conventional surface ships. Different simulation packages are available on a commercial basis. The accuracy of predictions obtained using these models are largely dependent on the level of accuracy in estimating the values of the parameters used in these models. Several methods are available for the determination of the values of these hydrodynamic parameters. However, each one of these existing methods has its own shortcomings. In this paper a new predictive method is outlined for the estimation of the hydrodynamic characteristics for a ship performing certain standard manoeuvres. The method uses neural networks technique to predict the hydrodynamic parameters of the ship. Parameters normally measured during standard manoeuvre e.g. surge, sway and yaw velocities and rudder angle, will be used as input to the predictive model. The first author has successfully used the same technique in predicting the transverse stability of a ship at sea.