Anti-windup for model-reference adaptive control schemes with rate-limits

This paper proposes an anti-windup like scheme for an LTI plant with rate-limits. The plant is controlled using a model-reference adaptive controller, making the anti-windup design problem highly nonlinear. It is assumed that the rate-limit is modelled as a first order feedback loop for which the state is unavailable, but that the bandwidth of this loop is known. The anti-windup scheme uses a “hedging” term and a “positive ” term. The structure of the problem makes the rate-limit case considerably more difficult than the magnitude limit case. Nevertheless it is proved that convergence of the system state to the ideal model can be accomplished under conditions similar to those found in anti-windup compensation for purely linear systems.