Estimation based multiple model iterative learning control

An iterative learning control (ILC) framework is developed which provides robust stability and performance bounds under the assumption that the true plant model belongs to a plant uncertainty set that is specified by the designer. A set of candidate plant models is defined comprising hypotheses of the `true´ plant model, and after each ILC trial the update used is chosen to correspond to the current best plant hypothesis from the observed history via an optimisation based estimation process. A comprehensive design procedure for the switched multiple model ILC system is presented which is applicable to a general class of ILC update.