Generalized predictive control for π-type rudder anti-pitching based on nonlinear adaptive intelligent model

For π-type rudder anti-pitching system, an iterative predictive control algorithm based on information fusion is proposed. Optimal estimation of co-state sequence and control sequence is obtained by the fusion of future reference trajectory in quadratic performance index function and soft constraint information together with hard constraint information in system function. By taking 442T surface ship as the research object, typical sea state hydrodynamic coefficients are obtained according to pool experiment results and strip theory, and an adaptive three-dimension nonlinear intelligence model is built according to real-time changing speed, sea situations and course angle. The Radial Basis Function Neural Network (RBF) is optimized by optimal stopping rule avoiding over learning. Finally, simulations on information fusion predictive control were conducted on the proposed adaptive intelligence model with sea colored noise. The results show that the proposed algorithm has good efficiency in adaptively controlling on ship´s longitudinal motion with all weather.