On the fault diagnosis of uncertain singularly perturbed systems

Abstract This work aims at the fault diagnosis of two-timescale, singularly perturbed systems (SPSs) and SPSs with a multimodelling structure. The approach of model-based fault detection and isolation using the dynamic parity space to generate singular perturbation parameter free residuals was used. A comparative study of diagnosis of the SPS is carried out considering, in the first case, the SPS as an uncertain system with parameter uncertainties and using, in the second case, the slow subsystem for fault detection and isolation. The results are applied to a power system.