Accounting for Uncertainty in Anti-Windup Synthesis

This paper describes an approach to synthesizing anti-windup compensators, which can improve the behaviour of systems subject to actuator saturation while also taking into account uncertainty in the system. The class of uncertainty considered is reasonably large and, moreover, is of the type often used in practice and often considered in linear robust control. The development of the ideas makes use of a 'decoupled' representation of an anti-windup scheme, which is useful for comparing the results from standard approaches to anti-windup compensation to those compensators obtained using the new robust approach. An interesting, but perhaps not surprising feature of these results is that the often-criticised internal model control (IMC) anti-windup solution emerges as an 'optimally' robust solution.