A Study On Intelligent Adaptive Control Of Ship Autopilots

The autopilots have played a very important role in ship maneuvering systems. It has been one of the most important research topics of ship's maneuvering system, so a lot of experts and scholars have worked hard on it. Research on autopilots of ships started in 20's of last century. Up to now, the four development stages have been experienced, such as mechanical, PID, adaptive and intelligent autopilots. At present, the intelligent autopilots are being studied by a lot of scientists, and there is room for improvement.The ship motion models are the basis of researching ship's motion maneuvering control. Aimed at the characteristics of ships, such as non-linearity, time-varying, uncertainty, the linearity and non-linearity mathematics models of ship motion are discussed in the dissertation firstly. Then, by combining fuzzy theory and neural network, a kind of fuzzy neural network model for simulating ship motion is presented. Moreover, the environment models of noise such as wind, ocean wave and current are given also.In order to satisfy the requirements for ship's autopilot in ship course-changing mode under various sea status and different environments, a reference model constructed by neural network is proposed. Based on the fuzzy control and neural network, an intelligent adaptive control algorithm is presented in the dissertation. In order to enhance its adaptive characteristics, the parameters of membership functions and fuzzy control rules are modified with neural network learning algorithm.Based on fuzzy cerebellar model articulation controller (FCMAC), a kind of intelligent adaptive control algorithm for ship steering control is proposed. In order to improving parameter optimization and control performance, by combining genetic algorithm (GA), forgetting factor algorithm and back-propagation (BP) algorithm, a kind of new intelligent adaptive control scheme for ship's course keeping control with hybrid learning algorithm is presented.The results of simulation have shown that the control algorithms and schemes proposed above are effective, and the controllers designed have good performances and adaptive abilities.Finally, in order to decrease the period of experimentation on the sea, to reduce the cost of experimentation, to shorten the development period, to keep security of experimentation on the sea, a new hybrid simulation system of ship motion is presented in the dissertation. The system has been designed and tested, and it provides a favorable test environment for the research of various algorithms and productions, such as autopilots and stabilization fins etc. The hybrid simulation system of ship motion is effective in practical applications.