Discrete inversion based FDI for sampled LPV systems

The paper investigates the design problem for detection and isolation of faults in linear parameter varying (LPV) systems by means of dynamic inversion where the system matrix depends affinely from the parameters. A method for the construction of the inverse, relying on the concept of parameter varying invariant subspaces and related concepts of classical geometrical system theory is presented. A discretization method is proposed for the filter implementation that exploits the structure of the original LPV model, formulated in the continuous time, while maintaining the stability of the zero dynamics of the original system. The proposed method is illustrated through an application example concerning the detection of aileron and rudder faults on a commercial aircraft.