Off-line robust fault diagnosis using the generalized structured singular value (Q1614397)

The paper presents an off-line robust model-based diagnosis design method for multivariable continuous-time uncertain systems. The designed system is activated after the judgement ``fault'' by an on-line fault detection module, to gain insight into the situation. The generalized structured singular value \(\mu_g\) based frameowrk and the model invalidation theory in the frequency domain by \textit{M. Newlin} and \textit{R. S. Smith} [IEEE Trans. Autom. Control 43, No. 7, 901--907 (1998; Zbl 0952.93124)] are applied to the fault diagnosis problem. Modeling and fault detectability issues are discussed. Sensitivity conditions for both a multiplicative fault and an additive fault are proved. A fault diagnosis scheme offering a novel step-by-step off-line fault identification and symptom-aided diagnostic is presented. A numerical example is supplied.