| Summary: | 碩士 === 銘傳大學 === 電子工程學系碩士班 === 97 === In the thesis, a Proportional-Derivative Fuzzy Cerebellar Model Arithmetic Controller (PDFCMAC) is considered to solve tracking problem of a class of nonlinear systems. The linguistic rule of Fuzzy theory is implemented to PDFCMAC, which if-part is membership function and then-part is CMAC. First, adaptation laws of the PDFCMAC are used to approximate an ideal controller, and then a compensated controller is employed to assure the system stability. Second, redesign adaptation laws for the proposed controller and the estimated error bound are concerned to attenuate the chattering control signal and promote the robustness of adaptation laws. The robustness of adaptive laws and compensated controller are respectively derived from the Lyapunov stability analysis, so that the system tracking ability and the error convergence can be guaranteed in the closed-loop system. Under the requirements of PDFCMAC stability and defined performance index, the Particle Swarm Optimization and fitness function are used to search a set of parameters of PDFCMAC, which is applied on electronic driver system, van der pol system and invereted pemdulum system. From the simulations, controller parameters searched by PSO has performance as well as redesign adaptive laws, and has smaller compensated control signal than redesign adaptive laws.
|
|---|
