Output stabilization of Takagi-Sugeno fuzzy systems (Q1973342)

The authors discuss a class of Takagi-Sugeno models governed by rules of the form \[ \text{-if } z_1\text{ is }M_{1i} \text{ and\dots and }z_p \text{ is }M_{pi} \text{ then }dx/dt= A_ix(t)+B_i u(t),\;y(t)=C_ix(t) \] (in the above \(x(t)\in \mathbb{R}^n\), \(u(t)\in \mathbb{R}^m\) and \(y(t)\in \mathbb{R}^q\), \(i=1, 2, \dots,r\); \(M_{1i},\dots,M_{pi}\) are fuzzy sets contained in the condition part of the rule). The problem under consideration deals with output stabilization in this class of models. Starting from observers formed for each rule (subsystem), sufficient conditions are provided for their asymptotic convergence. It is shown that a state feedback controller and an observer yield a stabilizing output feedback controller (assuming that the stabilizing property of the control and an asymptotic convergence of the observer are guaranteed by the Lyapunov method).