Study Of Several Key Problems On Fuzzy Control System

Although many successful applications were found in real world, frizzy controller, as a controller based on linguistic rules, doesn抰 have special magical power. The design of traditional controller was based on the mathematical model of plant. Fuzzy controller was often used where it is impossible or too expensive to obtain the mathematical model of plant. Since its beginning, fuzzy control system has been a field filled with disputation. Four aspects of fuzzy control systems were studied, namely, fuzzy modeling, the problems caused by the in-completeness of the rule base, fuzzy adaptive control system, linguistic stability analysis of fuzzy control system. The thesis is classified into six chapters and the contents are outlined as follows: In chapter 1, an introduction is first given to the background of fuzzy control. Then its development is reviewed. Finally, the main achievement and arrangement of the thesis are concluded. In chapter 2, an introduction is first given to the basic concepts of fuzzy control. Then its general principles and theoretical basis of fuzzy control are discussed. Finally, an discussion is given to its general structure and basic analysis methods. In chapter 3, fuzzy modeling methods are studied. It is classified into two sections: 1) Study the problems caused by sparse samples. Concept of localized trend analysis is used to provide a method to use the samples more effectively. 2) Provide a T-S type fuzzy system modeling method. Concept of local linearity measure is given, then a cluster method based on this concept is proposed, finally, the result obtained by clustering is used to form a T-S type fuzzy model. In chapter 4, a frizzy controller based on the distance between two fuzzy sets is proposed. Firstly, some definitions of distance between two fuzzy sets are given. These definitions are different from traditional one抯 for their dependence on the cardinal numbers and centroids of the two fuzzy sets. Then the method is used in designing a fuzzy controller to handle the problem cased by the fact that the input universe is not covered completely with the rule base. In chapter 5, RLS algorithm is used to form a fuzzy model of the plant being controlled. and then an adaptive designed method of fuzzy controller is proposed based on the established model. Advantages of the method proposed are generating fuzzy controller adaptively, the experience knowledge about plant, controller or both, can be used 4? effectively to improve the performance of the adaptive method. In chapter 6, a linguistic method to analysis the stability of fuzzy control system is given where a large number of linguistic information about plant and controller is available though the mathematical model of the plant is unknown. A method to classify system states according to being stable or not is proposed first, then the space of system states are decomposed according their convergence.