Constructive solutions to the anti-windup problem

Linear control synthesis tools are routinely employed in a variety of applications, typically yielding excellent local performance. However, the plant input is always amplitude limited due to saturation which often causes undesirable larger signal behavior. The anti-windup problem is to design a 'fix' to mitigate the degradation due to saturation without altering the small signal behavior. In this dissertation, we will consider multiple performance objectives and structural classifications associated with anti-windup compensation. Systematic anti-windup compensation synthesis algorithms will be given which utilize two of the most powerful, yet computationally tractable, controller synthesis tools: linear matrix inequality and model predictive control. Successful application of such anti-windup compensation schemes will be demonstrated via simulation and experiment.