Fault Diagnosis Using Covariance Assignment and Regional Pole Placement of T-S Fuzzy Observers

This work is concerned with a new approach for model-based fault detection and isolation (FDI) of flight control system (FCS). Considering nonlinearities and cross-coupling of dynamic flight control, a T-S fuzzy model has been employed to represent nonlinear FCS with actuator/sensor faults. By the principle of parallel distributed compensation (PDC), the T-S fuzzy observers described by a number of locally-linearized models can be designed to provide characteristic features for FDI like residuals in the actual region of flight operating points. And by using theories of covariance assignment and regional pole placement, the problem of the T-S fuzzy observers with desired requirements of steady residual covariance and regional poles can be transformed into a problem of linear convex optimization and parameters of each observer at different operating points are obtained by linear matrix inequality(LMI) technique. Finally an example of normal acceleration monitor with changed aerodynamics and random vertical gust input of FCS has been presented to demonstrate the obtained result and show its performances