Intelligent Modeling And Control For Ship Lateral Motion

Based many years of study, it has been found by many scholars that the incertitude of ship model widely affect the effect of antiroll method. The hydrodynamic coefficients of ship's vertical motion change a lot in different sea environment and ship sailing condition. However, the hydrodynamic coefficients set up by the experimental data only focus on the representative navigating speed, ocean condition and course. This model is inapplicable for the complex navigating condition and the real time control in reducing the yaw and roll.This paper introduced three methods to approach hydrodynamic coefficients nonlinear function in three-dimension. The three methods are least square method, with interpolation method, invariable radial basis function (RBF) neural network and self-organizing radial basis function (SORBF) neural network. Finally, this paper gives the comparison of the three methods. The results of much simulation indicate that the relative error of least square method is the largest of all. The precision of RBF and SORBF is almost same, the largest relative error of all the 18 hydrodynamic coefficients is less than 2%. The intelligent model is able to satisfy the real time control in working condition. In the simulation, the construction and its algorithm of former model is simpler than the later one, so need less training time. However, the later model can be used in the hydrodynamic coefficients modeling of all kinds of ship.Focus on the intelligent model, this paper discusses two methods to reduce yaw and roll. The two methods are LQG control with dynamic parameters and self-organizing control. Finally, the comparison of the two methods is given. The simulation indicates that the first method is able to reduce the mean square error of roll angle to 1.42°; the control effect is about 46%-74%; the mean square error of yawing angle reduce to 1.68°, although the yaw increase by 15.2%. The self-organizing control is able to reduce the mean square error of roll angle to 0.92°, and the mean square error of yaw angle to 0.74°, the yaw increase by 13.9%. The performance of intelligent control is a little better than LQG control.