Robustness of Indirect Adaptive Control Based on Pole Placement Design

Abstract

We study the robustness of an indirect adaptive control scheme based on pole placement design with respect to unmodeled dynamics, non linearities, time variations or to ill-modeled measurement disturbances. The known results about this problem show that classical adaptation mechanisms have to be modified. Here introducing a regularized normalized least squares algorithm with a projection, we state a boundedness property in presence of mismodeling quantified in terms of noise to signal ratio. However an extra condition about the controllability of the adapted model is required.