A new framework for iterative identification and control

In this paper we present a new framework for iterative modeling and control. We begin by describing the unknown process with an uncertain model whose parameterization depends on prior information, available control design tools and other modeling preferences. The next step is an iterative procedure for refining the uncertainty set via robust control based model invalidation and can be viewed as a systematic way of efficiently searching for a controller delivering a certain desired level of performance to the unknown process. As a result, either the performance goal will be met or the entire uncertainty set will be invalidated in accordance with our modeling and control method prejudice. An iterative scheme based on a fixed pole model structure and rank one mixed μ synthesis is described in detail and a specific example is used to illustrate the ideas