Anti-windup Compensation using a Decoupling Architecture

Abstract

This chapter describes the theory and application of a ‘decoupled’ approach to anti-windup compensation. One of the salient features of this approach is that the anti-windup problem can be posed in a manner which decouples the nominal linear closed-loop from the nonlinear stability problem associated with actuator saturation and the anti-windup compensator. This process also reveals a logical and intuitive performance criterion which one can optimise by means of linear matrix inequalities. Details of how the anti-windup problem can be solved using various forms of compensator are given, together with some robustness considerations. Initially, the problem is posed for globally asymptotically stable plants, although this condition is later relaxed to allow the consideration of plants with exponentially unstable modes. Finally, the results are demonstrated on a multivariable aircraft model and the benchmark inverted pendulum.