Design of fuzzy adaptive iterative learning control for nonaffine nonlinear discrete-time systems

This paper considers the design of iterative learning control for a class of nonaffine nonlinear discrete-time systems which can perform a repeated task over a finite time sequence. A new error model formulation is derived such that an unknown nonlinearity can be approximated by a fuzzy system with available input and output signals. The function approximation error and external random disturbance are then compensated by an auxiliary error with dead zone like design. The fuzzy consequent parameter vector and the boundary layer of dead zone are updated for the next iteration in order to guarantee the learning error convergence. It is shown that output error will converge to a region around zero whose size depends on the width of boundary layer if the learning gains of adaptive laws satisfy a sufficient condition.