Nonlinear modeling of a flight vehicle using fuzzy clustering
Ali Reza Mehrabian, Soheil Hashemi, Jafar Roshanian, Salman Mohagheghi · 2006
An automated procedure for gain-scheduling identification of nonlinear systems by means of fuzzy clustering is presented. Fuzzy clustering is one of well-established learning methods that can be used for extracting Tagaki-Sugeno (TS) fuzzy rule set. It is known that TS fuzzy systems apply soft gain- scheduling which makes them favorable for combining with classic gain-scheduling (GS) design method. In conventional GS design method, there exist two main difficulties which are: I) obtaining appropriate operating (also called set points or equilibrium points) for applying linearization and linear controller design for the nonlinear system, and II) interpolation route between obtained operating points. Fuzzy clustering is reported to solve both these problems simultaneously, that makes the GS a safe and reliable controller design method for nonlinear systems. The design and implementation of fuzzy clustering for a flight vehicle which has non-minimum phase zero and suffers from poor stability is reported to reveal advantages of the algorithm.