本論文主要針對Takagi-Sugeno 模糊控制器之設計提出一個以實數型基因法則為基礎同時結合LMI (Linear Matrix Inequalities) 求解技巧的新演算法，由於本演算法同時具備基因法則的最佳化搜尋特性與 LMI 求解技術之優點，故透過不同適應函數之制定，便能夠設計出符合各種系統特性與性能要求的控制器，以改良一般 LMI 設計方法的不足之處。本文係以自走車動態系統為例，藉由其軌跡收斂控制器的設計，以分析現有的 LMI 技術與本文所提之方法，並且透過實際的模擬，作一系列最詳盡的討論與比較。文末敘述如何研製一個自走車系統，並運用本文方法所設計的 T-S 模糊控制器進行軌跡收斂的實驗，以驗證其可行性與可應用性。 In this paper, we propose a new algorithm to design the Takagi- Sugeno fuzzy controller via Real-Valued Genetic Algorithm (GA) and LMI (Linear Matrix Inequalities) technique. Because this algorithm combines the advantages of GA and LMI, one can design the feedback gains of the controller according to characteristics and required Performances of each system. Hence, this method can improve the drawbacks of general LMI-based design ones. This thesis first discusses the design of the controller of the trajectory stabilization of a model car. Then we dilate on the process of design controller by two different ways and compare them by means of simulations. In order to show the feasibility and the applicability of this method, we explain how to devise and create a model car and verify it with the T-S fuzzy controller based on the GA and LMI.
