Robust iterative learning control for nonlinear system based on T-S model

A design method of robust iterative learning controller (RILC) is proposed for nonlinear system with repetitive actions in this paper. First, the Takagi and Sugeno (T-S) fuzzy model is employed to approximate a nonlinear system with repetitive actions, and thus global fuzzy system model is displayed as the form of uncertain systems. Next the robust iterative learning controller is designed by the T-S fuzzy model employed as a dynamic model of nonlinear system. The proposed controller is obtained by solving linear matrix inequalities (LMI), and both sufficient and essential of the convergence conditions are deducted by Lyapunov theory. The proposed controller needs not satisfy the condition of fully observable as well as we did before. A single inverted pendulum problem has been simulated with RILC and compared with iterative learning controller (ILC) and the results show that RILC performs better than ILC.