| Summary: | 碩士 === 國立臺北科技大學 === 電機工程系 === 107 === This thesis organizes the concepts of dynamic screening mechanism, self-organizing recurrent cerebellar model articulation controller and TSK fuzzy system architecture to fulfill an adaptive TSK Fuzzy Self-organizing Recurrent Cerebellar Model Articulation Controller with Dynamic Screening Mechanism (ATSKSORC-DSM).
The novel design concept of this controller is to add dynamic error screening mechanism, adopts cerebellar model controller, self-organizing recurrent cerebellar model controller architecture, TSK fuzzy rules and adaptive learning rules, in order to optimize the output response of controlled systems and improves the acceleration of the convergence in the transient state. Simultaneously, the CMAC, which is originally static and fixed number of associative memory layers, has the performance of dynamic memory and the ability to correct the number of associative memory layers. According to the layer decision-making mechanism, the number of associative memory layers will be adjusted. In addition, ATSKSORC takes the integral error function as input and introduces it into the self-organizing recurrent cerebellar model controller.
Furthermore, the compensating controller is designed to dispel to the errors between an ideal controller and the TSK fuzzy self-organized recurrent cerebellar model controller. To ensure the stability of the control system, the Lyapunov's stability theorem is applied to derive the adaptive learning laws of TSK parameters, recurrent weights, the mean parameters and the standard deviation of Gaussian function, respectively.
In this study, the adaptive TSK fuzzy self-organized recurrent cerebellar model controller with dynamic screening mechanism is used in chaotic systems as the synchronous controller and stabilizing controller.
In synchronizing and stabilizing the chaotic systems, experiments for comparisons among the conventional CMAC, FCMAC, and ATSKSORC are performed. The experimental results reveal the proposed ATSKSORC has better robustness and control performance in different simulation conditions.
|
|---|
