Variable structure adaptive observer approach for actuator fault detection and diagnosis in uncertain nonlinear systems
Proceedings of the American Control Conference
We construct a variable structure adaptive observer using only input and output to diagnose actuator faults by a learning method. Stability condition of the proposed observer is relaxed by introducing the expansions of nonlinearities into power series. The observer's gain matrix can be calculated easily. The robustness properties of the observer ensure that no false alarms are registered due to uncertainties and disturbances in the system.
Chen, Wen; Saif, Mehrdad; and Soh, Yeng Chai. (2000). Variable structure adaptive observer approach for actuator fault detection and diagnosis in uncertain nonlinear systems. Proceedings of the American Control Conference, 4, 2674-2678.