Font Size: a A A

Study Of Intelligent Predictive Control For Ship Autopilots

Posted on:2003-09-09Degree:DoctorType:Dissertation
Country:ChinaCandidate:C D ZangFull Text:PDF
GTID:1118360092466275Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The autopilots of ships have been the importance research area of ships steering control system for long time. It is closely related with sailing safety, energy saving and releasing burden on the pilots. A lot of experts and scholars in this area have worked hard on it. Research on autopilots of ships started from 20's of last century. The first generation of autopilots was of the mechanical type. As the development of control theory and electronics technique, the second generation and the third generation of autopilots appeared in the type of PID and adaptive schemes. Along with the development of neural networks, fuzzy system etc., so called the intelligence technique, the fourth generation of autopilots has been focused.The ship dynamic model is the foundation in the study on autopilots of ships. Three different kinds of ship dynamical model are proposed in the dissertation. The wind, current and sea wave disturbances can produce strong influence on ship sailing. Therefore, the wind, current and sea wave disturbances are modeled also.Generalized predictive control algorithm for autopilots of ships is discussed. This scheme adopts CARIMA model to predict output of the system. The generalized predictive control law for autopilots of ships can be derived based on rolling optimization of the performance index.Generalized predictive control algorithm for autopilots of ships based on neural fuzzy model is proposed. It uses neural fuzzy model as a predictive model. The fuzzy model adopts the T- S type structure. Its parameters are trained by using neural network. It can be used not only in linear systems, but also in nonlinear systems.Some schemes of the neural networks predictive control for autopilots of ships are discussed. A neural network predictive control based on neural networkmultistep prediction and a neural network predictive control for autopilots of ships based on fuzzy predictive model are proposed. Simulation results have shown that these schemes not only have the same performances as that of conventional predictive control, but also appear stronger robustness in nonlinear case.Finally, a fuzzy predictive control for autopilots of ships is proposed.The simulation results have shown that the proposed control schemes above have better performances than the conventional PID control scheme, and have stronger robustness than the adaptive schemes.
Keywords/Search Tags:ship autopilots, predictive control, neural networks control, fuzzy control
PDF Full Text Request
Related items