Robust sliding-mode observer-based multiple-fault diagnosis scheme

In this study, we simultaneously evaluate the multiple-fault diagnosis problem of a class of Lipschitz nonlinear systems with actuator and sensor faults and unknown input disturbances. A nonsingular system transformation is used to transform the original system into two subsystems for multiple-fault diagnosis: subsystems 1 and 2. At the system level, two robust sliding-mode observers (RSMOs) are proposed. An RSMO is designed for subsystem 1 to detect actuator faults subjected to unknown input disturbances, and another RSMO is designed for subsystem 2 to detect sensor faults subjected to actuator faults. At the component level, a bank of RSMOs is proposed to detect and isolate actuators (sensors) with faults using a dedicated observer scheme. The reachability of RSMOs is comprehensively investigated in the estimation error space. Accordingly, the proposed observer parameters are designed as an optimization problem and solved using the linear matrix inequality (LMI) optimization technique. The effectiveness of the proposed multiple-fault diagnosis scheme was validated through simulations of a modified seventh-order aircraft system.