Further results on switched control of linear systems with constraints

In a previous paper, we proposed a supervisory control system to globally regulate to zero the state of a very poorly modeled, open-loop unstable but not exponentially unstable, linear process in the presence of input constraints. The process to control was unknown but assumed to belong to a finite family P of nominal models. In this paper, the analysis is extended to the case in which P is not finite and is a closed, bounded subset of a real, finite-dimensional, normed linear space. In this analysis, the property of the multiestimator/multicontroller of being "robustly" integral input-to state stable is exploited.