Iterative design of robust generic model controllers for industrial processes

Abstract

The use of robust output-rate reference systems is presented in this paper as a new approach of accommodating process uncertainty within the GMC design framework. Three different design methods for the robust GMC reference system, namely the optimal, simple and iterative designs, are proposed. The design methods are derived from the application of the ν-gap metric and the H ∞ loop shaping method within the GMC design framework. The optimal design is achieved by solving an H ∞ optimization while the simple design is derived after analyzing the low and high loop gains of the optimally robust GMC reference system at different frequency ranges. The simple design assumes that the robust GMC reference system is parameterized in terms of a single variable, β, and requires an estimate of the process uncertainty. In the case that we have a conservative estimate of the process uncertainty, a data-driven optimizer is optionally applied to iteratively adjust the parameter β for an improved closed-loop performance. As applied to two simulated industrial case studies, the robust GMC controllers outperform the standard GMC controller in spite of large model uncertainty.