Interval estimation for continuous-time linear switched systems via ellipsoidal analysis

This paper proposes a two-step interval estimation approach for continuous-time linear switched systems with unknown but bounded uncertainties. First, a robust observer design method is proposed to obtain point estimation and attenuate the effects of the system uncertainties. Second, based on a system split method, an ellipsoidal analysis method is introduced to provide the smallest enclosures of the estimation error. By merging the interval evaluation of the estimation error with the point estimate provided by the robust observer, the proposed method allows one to obtain a tight interval estimation of the actual state vector with less restrictive and simple design conditions. The proposed method solves the interval estimation problem for switched systems in two cases: switched systems under an arbitrary switching signal and switched systems under an average dwell time switching condition. In terms of accuracy, comparison studies show that the proposed method outperforms some existing methods in the literature.