Robust fault detection in uncertain polytopic linear systems.

Abstract

This paper presents a novel approach for robust fault detection and isolation filter design, based on modifiedH2-H_ performance conditions, which are described as LMIs. In order to synthesize the residual generation scheme, some theoretical results are shown for systems with perturbation signals and subjected to parametric uncertainty. The uncertainty parameters are supposed to belong to a polytope. The extended H2-H_ conditions are obtained by means of the well known projection lemma. Fault detection and isolation are done by using a filters bank (i.e. multifiltering) based on Luenberger”™s observer and one filter is obtained for each fault. Performance of the proposed synthesis technique is illustrated by a numerical example.

Keywords

• fault diagnosis
• robust control
• linear matrix inequalities (LMI)
• H2-H norms