An approach to adaptive control of continuous-time systems using multiple parallel predictive models

Abstract

An approach to the adaptive control o continuous-time systems using multiple parallel predictive models is presented. It is motivated by the generalized predictive control method in discrete time of D. W. Clarke et al. (Automatica, vol. 23, no. 2, pp. 137-148, 1987). A bank of N parallel predictive models, each one applicable to a plant model of different relative degree, is used to develop an adaptive control strategy that requires only knowledge of an upper bound on the order of the plant but does not require exact knowledge of the relative degree. The adaptive controller is also applicable to nonminimum-phase systems.