Takagi-Sugeno fuzzy modeling and control: unmodeled dynamics and robustness

The Takagi-Sugeno (TS) fuzzy modeling approach has been very successful in approximating complex dynamics and hence has offered a basis for analysis and design in many control applications where conventional modeling is impossible. At present, a drawback of the TS fuzzy model is the high dimension of its information space, which, in turn, necessitates a large number of local linear models to cover a certain region of such space. This article shows that the dimension of the information space is reducible only at the expense of modeling error. The article also suggests a robust controller that tolerates a certain amount of such modeling error.