Input-to-state Stability And Model Predictive Control Of Switched Systems

The switched system is widespread in the aerospace, automotive industry and the bio-engineering felds. Thus, the investigation of switched systems has theoretical signifcanceand important application. This dissertation gives a study on the input-to-state stabilityand model predictive of switched systems:1. The input-to-state stability of two classes of switched systems is discussed. In thepresence of feedback delay and actuator error, we utilize a common Lyapunov-Krasovskiifunctional based on the original undelayed dynamics to establish an input-to-state prac-tical stability framework for uncertain switched nonlinear systems. Based on dissipationinequalities, the input-to-state stability (ISS) of switched polynomial nonlinear systems isconsidered. Sufcient conditions are developed to ensure the ISS of switched polynomialnonlinear systems under arbitrary switching signal. A sum of squares (SOS) decompo-sition approach is provided to search for a ISS common Lyapunov function for switchedpolynomial nonlinear systems.2. Observer-based fault-tolerant control of continue and discrete-time switched sys-tems is investigated. A robust estimator is proposed to simultaneously estimate the stateof the system and the actuator fault. Based on the estimate of fault and the states, anobserver-based fault tolerant controller is designed to guarantee the input-to-state stabil-ity of the closed-loop switched system under average dwell time. Sufcient conditions forthe existence of the fault tolerant controller are formulated in terms of linear matrix in-equality. A switching signal depending on state is designed to guarantee the input-to-statestability of the closed-loop switched system.3. The problem of discrete-time switched systems with state delays and missing mea-surements is considered. Based on average dwell time method, sufcient conditions forthe existence of output feedback controller is designed to guarantee the input-to-state sta-bility of the closed-loop switched system under synchronous switching and asynchronousswitching. Finally, an output feedback controller is designed for discrete-time switchedsystems with varying delays and missing measurements.4. The problem of model predictive control and invertibility of switched systemsis presented. Using Lyapunov-based model predictive control, a distributed model pre-dictive control architecture is designed for a class of switched nonlinear systems subjectto communication disruptions and delayed measurements and required to satisfy a pre-scribed switching sequence. Sufcient conditions for guaranteeing ultimate boundednessof the closed-loop switched nonlinear systems are presented. In view of geometric terms,necessary and sufcient conditions for the invertibility of switched systems are given.