Filtering and fault estimation of descriptor switched systems

In this paper, the problems of state and fault estimation are addressed for a class of switched descriptor systems subject to Lipschitz nonlinearities and unknown inputs (UI). The UI appear both on the dynamic and on the measurement equations. Two problems are addressed by L2-gain minimization with the use of switched Lyapunov functions and formulated by LMI. First, a functional observer for switched Lipschitz nonlinear descriptor system is proposed for robust state estimation. Second, fault estimation is performed by filtering the output estimation error, as usually done in the residual generation framework. Moreover, frequency weighting functions can be used to shape the response to the fault and thus improve the estimation.