Adaptive predictive control of time-varying NARMA systems with exact output tracking

In this paper, adaptive control aiming at asymptotic tracking is investigated for a class of NARMA (nonlinear auto regressive moving average) systems with periodic time-varying parameters, unknown time delay and nonparametric nonlinear uncertainties. A time-varying dead zone with diminishing threshold is designed in the parameter estimator to deal with both parametric and nonparametric uncertainties. The effect of the nonparametric uncertainties as well as unknown time delay has been compensated for by the developed nearest-neighbor previous instant estimation technique. The closed loop stability and asymptotic tracking performance have been rigorously established by theoretic proof.