State Estimation of Networked Switched Systems via Event-Triggered Zonotopes

Abstract

In this article, the state estimation issue is investigated for networked switched systems via an event-triggered zonotope method, and the general asynchronism is considered between subsystems and observers. First, the event-triggered zonotope is proposed to enclose the system state, where the observer gain only updates with the switching signal at triggering instants. The restriction on the switching signal is removed such that system switchings can happen more than once over each interevent interval. Then, specific iteration procedures are provided for the generator matrix in synchronous and asynchronous intervals. Solvable conditions are proposed to jointly optimize the event-triggered parameters and observer gains by constructing observer-mode-dependent radius functions. The boundedness and ${\mathcal {L}}_{\infty }$ performance are analyzed for the radius of the zonotope under restricted switchings. Finally, an implementation to the H-bridge rectifier is presented, which validates the merits of the proposed method.