Event-based robust state estimator for linear time-varying system with uncertain observations and randomly occurring uncertainties

Abstract

This paper is concerned with the event-based robust state estimation problem for discrete time-varying system with uncertain observations and randomly occurring uncertainties. Norm-bounded uncertainty occurring in a random way is introduced in the state matrix. A Bernoulli distributed random variable is employed to characterize uncertain observations phenomenon. To efficiently utilize the network resources, an event-triggered scheme is adopted here to transmit the measurements to the estimator. The purpose of this paper is to propose two-stage event-based estimators including predictor and filter such that, in the presence of uncertain observations, randomly occurring uncertainties and event-based scheme, upper bounds on the estimation error covariance can be established and estimator׳s parameters are obtained. By minimizing the trace of estimation error covariances, estimator׳s parameters are derived, which are the solutions to discrete Riccati difference equations. The proposed estimation algorithm, expressed in a recursive form, involves the occurrence probabilities of uncertain observations and randomly occurring uncertainties, the threshold and an event indictor variable caused by the event-triggered scheme. Finally, numerical simulation results are provided to demonstrate the effectiveness of the developed estimators.