Adaptive tracking control of switched linear systems using mode-dependent average dwell time

Abstract

This paper studies model reference adaptive control for switched linear systems with large parametric uncertainties. An aggregate leakage approach is proposed to develop a novel adaptive law, which overcomes the state-of-the-art assumption of knowing the upper and lower bounds of the parameter uncertainty. In addition, a switching law is developed based on mode-dependent average dwell time scheme, which exploits the information of the known reference model for every subsystem, i.e., average dwell time is realized in a subsystem sense. Based on the proposed time-constraint scheme, switching signals that are less conservative than those based on dwell time and average dwell time can be designed. Global uniform ultimate boundedness of the closed-loop adaptive switched system is guaranteed. Furthermore, the tracking error is shown to be upper bounded and also an ultimate bound is presented. Simulations using NASA GTM aircraft illustrate the proposed method.