A suboptimal learning control scheme for non-linear systems with time-varying parametric uncertainties

In this paper, learning control is integrated with non-linear optimal contr ol to enhance control performance of a class of non-linear systems with tim e-varying parametric uncertainties. A suboptimal control strategy based on a control Lyapunov function (CLF) and Sontag's formula provides suboptimal performance as well as stability along the time horizon for the nominal par t of the non-linear dynamic system. The proposed learning mechanism learns the unknown time-varying parametric uncertainties so as to eliminate uncert ain effects. System information both in time horizon and learning repetitio n horizon are incorporated in a composite energy function (CEF). The propos ed control scheme achieves asymptotic convergence along the learning repeti tion horizon and boundedness and pointwise convergence of the tracking erro r (perfect tracking performance) along the time horizon. Copyright (C) 2001 John Wiley & Sons, Ltd.