Robustness of model reference adaptive controllers: an input-output approach

A general input-output approach is employed to study the direct model reference adaptive control problem in the presence of parametric and unstructured uncertainty. A sufficient condition for the boundedness of the closed-loop signals is developed, relating an appropriately defined size of the unstructured uncertainty to the size of the parametric uncertainty. This condition involves the intuitively appealing gains of various sensitivity operators and allows for the study of the effect of the design parameters on the boundedness properties of the closed-loop system. The results of this work indicate a direction towards a unified and systematic theory in adaptive control.<
>