Robust identification for output tracking in systems with unknown nonlinearities

Abstract

We study the problem of identification and output tracking for uncertain strict-feedback systems where the unknown part of the system depends only on the output. The unknown nonlinear terms in the system description are not linearly parametrized, but it is assumed that the optimal parameters that characterize a neural net based approximation to the nonlinearities lie within a known compact set. The proposed controllers, with full state and derivative information, guarantee the boundedness of all signals in the closed-loop system. Also, at the expense of increased control effort, the output tracking error can be driven to an arbitrarily small neighborhood of the origin arbitrarily fast. Moreover, by specifying an appropriate reference signal, the satisfaction of a relevant persistency of excitation condition is ensured, and hence an adequate identification of the uncertain system is achieved.