Fault detection for time-delay systems within robust residual based on finite memory observers - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 1998

Fault detection for time-delay systems within robust residual based on finite memory observers

José Ragot
CID

Résumé

Fault detection is usually achieved within residuals generation (based on analytical redundancy) and evaluation. Such methods need the knowledge of a system model and therefore have to be robust to disturbances such as model uncertainties and measurement errors. In order to free oneself from the drawbacks of classic methods, as the observer performance degradation, the concept of finite memory observer has been introduced. This paper gives a short summary of the observer structure and some of its properties to further focus on the choice of the residual. In general the estimation error is taken as the residual but another residual is defined here. This residual is the difference of two state estimations at the same instant but based on two overlapping observation horizons. The relevance of this paper is to prove that this special residual is more robust with respect to some model uncertainties than the estimation error if the fa let detection problem is tackled. Finally, the proposed method is applied to detect sensor faults for time-delay systems. The relevance of this paper is that the observer is based on a simplified model of the time-delay system that does not take into account the delayed states. The example given is a simulated one.
Fichier non déposé

Dates et versions

hal-01002798 , version 1 (06-06-2014)

Identifiants

  • HAL Id : hal-01002798 , version 1

Citer

Walter Nuninger, Frédéric Kratz, José Ragot. Fault detection for time-delay systems within robust residual based on finite memory observers. Workshop on on-line-fault detection and supervision in the chemical process industries, Jun 1998, Lyon, France. pp.CDROM. ⟨hal-01002798⟩
66 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More