Robust diagnosis of nonlinear systems with structured uncertainties via the T-S fuzzy UIFDO

For linear systems with structured uncertainties, the unknown input fault detection observer (UIFDO) provides a robust diagnostic approach by decoupling the fault signals from the unknown input term. However, in real applications, detection performance of the UIFDO degrades seriously as nonlinearities become significant in a wide dynamic operating range. In this paper, a novel UIFDO design and the Takagi-Sugeno (T-S) fuzzy model are combined to provide a T-S fuzzy UIFDO which effectively deals with nonlinear fault detection problems under structured uncertainties and input disturbance. Moreover, by applying the linear matrix inequality (LMT) technique, linear observer gains for each rule of the T-S fuzzy UIFDO can be obtained to guarantee the global stability of the system. Finally, a nonlinear benchmark example is provided to illustrate the proposed design procedures