Design of a Robust Polytopic Unknown Input Observer for FDI: Application for Systems described by a Multi-Model Representation

The main goal of this paper consists in fault diagnosis of systems represented by a multi-model approach. Partial knowledge of the system representation around operating points must allow to obtain its dynamic behaviour under modeling errors consideration. The goal of this paper is to decouple modeling errors through a wide operating range by a dedicated polytopic Unknown Input Observer (UIO). The robust polytopic observer allows an optimization of modeling errors matrix distribution and permits to develop a fault diagnosis strategy on a wide operating range. Polytopic observer stability is guaranteed by pole assignment established through Linear Matrix Inequality (LMI). The effectiveness and performances of the method are illustrated in simulation considering a CSTR.