Sensor fault detection and isolation using multiple robust filters for linear systems with time-varying parameter uncertainty and error variance constraints

In this paper, a robust sensor fault detection and isolation (FDI) strategy is proposed by means of the multiple model (MM)-based scheme. The proposed approach is composed of robust Kalman filters (RKF) with error variance constraints that are designed for a linear discrete-time system with parameter uncertainties affecting all the system matrices. The robust filter parameters are designed by solving two algebraic Riccati equations expressed in linear matrix inequality feasibility conditions. The goal of this multiobjective problem is to design a robust filter which is not affected by system perturbations and satisfies the performance requirements including an asymptotically stable filtering process as well as individually bounded estimation error variances with predefined values. The proposed multiple RKFs are used in the MM-based strategy to detect and isolate sensor bias faults having different severities. Finally, an illustrative numerical example is given to demonstrate the robustness and the estimation accuracy levels of our proposed FDI scheme as compared with a standard linear Kalman filter-based FDI method.