Design and Application of Adaptive Fuzzy Iterative Learning Controller Using Genetic Algorithms

• Main Authors: Ryan, 吳宗庭
• Other Authors: Chiang-Ju Chien
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/07024167206211353334

碩士 === 華梵大學 === 電子工程學系碩士班 === 97 === The learning gain of iterative learning control system has a close relationship to the convergence of learning error. In this thesis, we study the learning gain design approach for discrete iterative learning controller and propose a design structure to improve the learning convergent speed. The first part of this thesis is to apply the fuzzy logic system for the design of learning gain. The learning gain can be adaptively tuned according to the magnitude and the change of learning error. Speed-type and position-type approaches are presented for the implementation of adaptive fuzzy learning gain. However, the optimal design of the fuzzy logic system is hard to achieve because the consequent part is in general chosen by trial and error. Hence, in the second part of this thesis, a genetic algorithm is then introduced to search for the optimal parameters of the adaptive fuzzy learning gain. Furthermore, a modified version of genetic algorithm is also presented in order to improve the search speed of the optimal parameters. Finally, a numerical example is used for computer simulation to demonstrate the learning performance for the genetic algorithms based adaptive fuzzy iterative learning controller. Simulation results show that the consequent parameters can be determined no matter the speed-type or position-type fuzzy learning gain is used. Based on the optimal parameters given by the proposed genetic algorithm, the convergent speed and learning performance can be effectively improved.