TSK模糊控制器之多階化設計

Abstract

摘要 本研究將提出一個TSK Model (Takagi-Sugeno-Kang Model)之多階模糊控制器(Multistage Fuzzy Logic Controller , MSFLC)的設計方法來控制大規模且複雜的系統。此主要目的是在於減少控制器的模糊規則數，使規則的數目僅依據輸出/輸入變數和歸屬函數(Membership Function)的個數而增加，而其增加方式是以二次式比例增加。 多階模糊控制器的三種調整參數包含：規則庫、輸出/輸入變數歸屬函數和調整因子(Scaling Factor)，而這些參數是透過實數編碼之遺傳演算法則來搜尋最佳的解。 此研究提供一種有系統的方法來設計多階模糊控制器；而此多階模糊控制器設計主要分為二元數及非對稱性樹兩種架構。最後，將藉由倒單擺系統(Inverted pendulum system)與球軸系統(Ball and Beam System)兩個受控系統來模擬驗證此控制器的性能。其模擬結果顯示此多階模糊控制器可行，並且確實使用較少的模糊規則數。

關鍵字：多階模糊控制器、遺傳演算法則、倒單擺系統、球軸系統

Abstract This paper proposes TSK (Takagi-Sugeno-Kang) Model approach to design a multistage fuzzy logic controller for large-scale and complex control systems. The main purpose of this paper is to decrease the large number of rules by using multistage fuzzy logic controller, and adopt the real coding genetic algorithm method to design the parameters on the multistage fuzzy controller. The fuzzy rule number of the proposed approach increase only quadratic with the number of inputs and membership function. There are three kinds of parameters on multistage fuzzy logic controller. It includes the rule base, input/output variables of membership function and scaling factors. They are all designed by real coding Genetic Algorithm. The proposed approach provides a systematic way to design multistage fuzzy logic controller, and these parameters design can be combined with two kinds of framework (Skew tree and Binary tree). Therefore, there are two kinds of multistage fuzzy logic controllers, and we compared their performance in pendulum-cart system and Ball and Beam system. The simulation results show that the multistage fuzzy logic controller can work and use less fuzzy rules.

Keywords：TSK Model、Multistage Fuzzy Logic Controller、Genetic Algorithm、Inverted pendulum System、Ball and Beam System