Inference of candidate loop performance and data filtering for switching supervisory control

The paper studies the problem of inferring the performance of a linear feedback-loop consisting of an uncertain plant and a candidate controller from data taken from the same plant possibly driven by a different controller. In such a context, a convenient tool to work with is a quantity called normalized discrepancy. This is a quadratic measure of mismatch between the loop made up by the unknown plant in feedback with the candidate controller and the nominal "tuned-loop" related to the same candidate controller. It is shown that discrepancy can be in principle obtained by resorting to the concept of a virtual reference, and conveniently computed in real-time by suitably filtering an output prediction error. The latter result is of relevant practical value for on-line implementation and of paramount importance in switching supervisory control of uncertain plants, particularly in the case of a coarse candidate model distribution. (C) 2001 Elsevier Science Ltd. ALl rights reserved.