Robust fusion filter for multisensor descriptor system with uncertain‐variance noises and packet dropout

Abstract For the multisensor linear stochastic descriptor system with uncertain‐variance noises and packet dropout, the standard state space model of the new reduced‐order augmented state is obtained based on the singular value decomposition (SVD) approach and augmented state approach. By applying the min‐max robust estimation principle and Kalman filtering theory, the local robust time‐varying Kalman filters of the reduced‐order augmented state and fictitious process noise are proposed. Then, the local robust time‐varying filters of the original descriptor system are presented according to the relation of the reduced‐order augmented state and the original state. And the distributed and covariance intersection (CI) fusion robust time‐varying filters are presented, and the actual and conservative filtering error variances are obtained. Further, the robustness of the local and fused robust filters are proved by applying Lyapunov equation approach, and the accuracy relation between them is analyzed. A simulation example applied to a circuit system shows its effectiveness and applicability.