Model Reference Adaptive Controllers and Stochastic Self-Tuning Regulators—A Unified Approach

Abstract

The control objectives for Model Reference Adaptive Control Systems (MRAC) are specified in terms of a reference model. The control objectives for Stochastic Self-Tuning Regulators (S-STURE) are specified in terms of an ARMA model. These two types of adaptive control systems share a common characteristic (duality). This duality extends the one existing in the linear case with known parameters between modal control (pole-placement) and minimum variance control. Based on the duality properties, a unified presentation of the two techniques is possible; the S-STURE being interpreted as a stochastic MRAC. The underlying concepts and the structure of these adaptive control systems are described. The available theoretical results concerning the analysis of these systems are briefly reviewed. The primary domain of application of these techniques are indicated. An overview of the unsolved problems concludes the paper.