Skip to Main content Skip to Navigation

Génération et évaluation de résidus pour le diagnostic de systèmes incertains : approche fréquentielle

Abstract : This thesis deals with the improvement of residual generation and evaluation approaches for diagnosis of uncertain systems in the particular case where parametric and unknown inputs are considered. The first part of this thesis presents robust diagnosis for uncertain systems. Major approaches proposed in literature are studied and the relevilience of the use of frequency methods for strongly uncertain systems is shown. In the second part, a residual generator based on a nominal model of the uncertain system is developed in order to improve fault sensitivity of residuals apd also to attenuate at best the effect of different disturb\lnces due to a natural process working and consequently, not to be detected. The residual generator is based on H_/H_ state observers'which frequency properties allows to consider fault isolation constraints respect to disturbances. The LMI optimisation techniques (Linear Matrix Inequalities) are then used. The last part studies parametric uncertainty effect on residual evaluation in such a way that an optimal threshold is determined to detect the smaller fault as possible and remaining insensitive to the largest model error. The considered parametric uncertainties are represented by a polytopic form. The proposed method is based on the use of Kharitonov polynomials and on the computation of bounding sets in the frequency domain. The evaluation is achieved by a RMS type of criterion which is formulated to link the spectral properties of detection signals with their evolution in the time domain.
Document type :
File URL :
Complete list of metadatas
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 10:42:28 AM
Last modification on : Monday, December 14, 2020 - 1:39:37 PM


  • HAL Id : tel-01746674, version 1



Florence Louis Rambeaux. Génération et évaluation de résidus pour le diagnostic de systèmes incertains : approche fréquentielle. Autre. Université Henri Poincaré - Nancy 1, 2001. Français. ⟨NNT : 2001NAN10018⟩. ⟨tel-01746674⟩



Record views