Uncertainty model analysis for MISO systems Set-Membership identification

Abstract

A geometric analysis of different structured uncertainty model sets for linear, time-invariant, multi-input single-output systems, compatible with Set-Membership identification methods, is presented. Additive uncertainty is assumed, whose size is measured in H∞ norm and which is described by two stable transfer matrices acting as weights and a norm-bounded uncertainty matrix. Two uncertainty models are obtained, the first one is independent for each input and can be obtained by Set-Membership classical methods, and the second one is a coupled model whose volume can be smaller, although its construction implies a nonconvex optimization problem.