Fault detection and diagnosis based on parameter set estimation for measurements corrupted by bounded noise

The objective of this paper is the Fault Detection and Diagnosis (FDD) of multiple abrupt parameter variations in an inherently time-invariant system based on Parameter Set Estimation. Under the a priori knowledge of the bounds of the noise that corrupts the measurements, it is proven that the parameter uncertainty is also bounded. The goal of the PSE is to compute at every time instant the orthotope that represents the bounded parameter uncertainty and within which the nominal parameter vector resides. A fault detection criterion is activated at the time instant that the computed orthotope is an empty set, while a 'backward-in-time' procedure is proposed for a more accurate estimation of the time of fault occurrence. Finally, the fault diagnosis continues with the isolation of the faulty components and the determination of the size of parameter deviation. Simulations studies are used to verify the efficiency of the suggested strategy for the case of multiple faults in a micro-electrostatic actuator.