Set-membership functional diagnosability: Definitions and analysis

Carine Jauberthie (1) , Nathalie Verdière (2) , Louise Travé-Massuyès (1)

1 LAAS-DISCO - Équipe DIagnostic, Supervision et COnduite
2 LMAH - Laboratoire de Mathématiques Appliquées du Havre

This paper introduces the concept of functional diagnosability for nonlinear dynamical uncertain models. Set-membership functional diagnosability is based on the notion of functional signature introduced in [11] extended to the set-membership framework. The link between functional diagnosability and set-membership functional diagnosability is established. Contrary to classical definitions, the study of functional diagnosability highlights some of the residual properties related to the fault acting on the system in time. Set-membership functional diagnosability extends this concept to bounded faults, i.e. faults whose magnitude is unknown but belongs to a bounded set. Numerical simulations for a system based on water tanks illustrate the potential of set-membership functional diagnosability.

Domaines: Systèmes dynamiques [math.DS]; Analyse numérique [math.NA]; Optimisation et contrôle [math.OC]; Traitement du signal et de l'image [eess.SP]; Automatique / Robotique; Santé publique et épidémiologie

Carine Jauberthie, Nathalie Verdière, Louise Travé-Massuyès. Set-membership functional diagnosability: Definitions and analysis. 2016 Conference on Control and Fault-Tolerant Systems (SysTol), Sep 2016, Barcelona, Spain. ⟨10.1109/SYSTOL.2016.7739794⟩. ⟨hal-01739797⟩