Robust fault diagnosis for linear descriptor systems using proportional integral observers

Benoît Marx (1) , Damien Koenig (1) , Didier Georges (1)

1 LAG - Laboratoire d'automatique de Grenoble

This paper presents the design of a proportionalintegral observer for descriptor systems subject to faults and unknown inputs. The observer is synthesized to minimize the influence of unknown inputs on the estimation. Weighting transfer is introduced to shape the sensitivity of the estimation to the unknown inputs. Particular attention is paid to fault diagnosis objective. The proposed method is based on the solution of LMI and guarantees the estimation of the states and faults to be robust face to unknown inputs. A numerical example is included.

Domaines: Automatique

Benoît Marx, Damien Koenig, Didier Georges. Robust fault diagnosis for linear descriptor systems using proportional integral observers. Conference on Decision and Control,, 2003, Hawaii, United States. ⟨10.1109/CDC.2003.1272605⟩. ⟨hal-00232850⟩