Subspace method aided data-driven design of observer based fault detection systems

Authors:	Ding Steven, University of Duisburg-Essen, Germany Zhang Ping, University of Duisburg-Essen, Germany Huang Biao, University of Alberta, Canada Ding Eve, University of Applied Science Gelsenkirchen, Germany
Topic:	6.4 Safeprocess
Session:	Signal Based Fault Detection and Isolation
Keywords:	Fault detection; fault isolation; observer based fault detection; subspace methods.

Abstract

This paper deals with the data-driven design of observer based fault detection and isolation (FDI) systems. The basic idea is to identify parity space and the related parameter matrices, instead of a state space model of the process under consideration, directly from test data. The proposed method can be used for the data-driven design of parity space, observer and kalman filter based FDI systems.