Study On Ship Motion Control Systems Based On Neural-Fuzzy Networks

The performance of autopilot is of great importance for a ship to navigate safely and economically. But because of the nonlinearity, time varying aspects and uncertainties of ship steering, PID control algorithm is difficult to adapt the changing situation, the conventional adaptive control methods are also not effective enough because it's difficult to find the accurate mathematical models. Since fuzzy control is very suitable for those complicated control objects whose accurate models are unavailable, and neural networks are strongly capable of learning ability and error tolerance, it's an important measure to apply the intelligent control theory, such as fuzzy control and neural networks, to the case of ship motion control, promising to solve the problem mentioned above. Thus, this dissertation carries out the study on neural-fuzzy networks with application to ship control. The main contributions of the thesis are summarized as follows.1. The mathematical model of ship motion with three degrees of freedom is established from the view of dynamics, where the model of rudder control machine and the disturbances caused by wind, wave and flow have been taken into consideration. Further more, using the S function of Matlab software, an Abkowitz nonlinear Simulink model is established, with which ship maneuvering test and the study of ship motion control law can be carried out in Simulink environment conveniently.2. The principles of course keeping and tracking control are studied and the applications of fuzzy control theory to the ship course-keeping control are investigated. According to the the different characteristics of ship maneuvering in stages of course changing and course keeping, fuzzy controllers for course changing, course keeping...