چكيده به لاتين
Abstract:
This thesis considers the design of robust H∞ filters for both continuous-time and discrete-time uncertain descriptor systems. Two types of uncertainties, norm bounded uncertainties and convex polytope uncertainties, are included. For discrete-time descriptor system full-order and reduced-order H∞ filters has been introduced. Also, resilient H∞ filters for discrete-time uncertain descriptor system with norm bounded uncertainties has been introduced. Moreover, the design of robust H2 and mixed H2/H∞ full-order filters for uncertain continuous- time descriptor system has been considered. The uncertainties are norm-bounded or polytope types. The proposed conditions for designing all filters are extracted without decomposing the original system matrices and are expressed in terms of convex and strict linear matrix equalities (LMI). The parameters of filters are extracted from these solvability conditions. The proposed necessary and sufficient conditions and also the proposed new sufficient conditions are able to obtain smaller attenuation levels and consequently can give rise to less conservative design than existing methods. A numerical example with simulation results is given, for each introduced method, to illustrate the effectiveness of the method.
Keywords: linear systems, robust filtering, strict LMI, uncertain descriptor systems
