Robust time-varying Kalman predictor for descriptor system with random one-step measurement delay and packet dropout

Abstract

This paper investigates a Kalman predictor with robustness is applied to descriptor systems with uncertainty of noises variances, random packet loss and measurement delay within one-step is addressed. First of all, taking singular value decomposition (SVD) method to reduce the original descriptor system, then the information which has been lost due to the poor network is compensated by enlarging the states and leading in fictitious noises, and then the system with a singular matrix is transformed into a standard system. With the use of robust minimax estimation principle, a robust predictor under the case of poor communication network is proposed. Under the condition that noise variances is unknown, the actual filtering error variance could be guaranteed within a minimal scope, and take Lyapunov equation to prove the robustness. To verify the conclusion presented in this paper, a simulation example has been shown at the end.