Robust adaptive techniques for sensor fault detection and diagnosis

In this note, a framework for automated detection and diagnosis of a class of sensor faults in linear systems is developed. This framework, which addresses both abrupt and incipient faults, utilizes a nonlinear adaptive detection observer along with a parameter estimator for failure assessment. The sensor failure is modeled either as an additive perturbation or as a multiplicative perturbation of the sensor signal that has magnitude described by a time varying function. Robust adaptive schemes are introduced to account for modeling errors that affect the diagnosis process and may cause false alarms. Further extensions from standard adaptive control techniques are applied to guarantee convergence of certain signals to residual sets. Numerical studies for mechanical systems are presented to demonstrate the applicability of the proposed automated diagnostic scheme.