A hierarchical approach for discrete-event model identification incorporating expert knowledge - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

A hierarchical approach for discrete-event model identification incorporating expert knowledge

Résumé

Recently, a new technique for identification of Discrete-Event Systems (DES) with the aim of fault detection has been proposed in the literature, where a model is obtained from the observation of the fault-free system behavior. In some cases, the system may execute different tasks in a sequential order to perform the complete operation of the system. In these cases, the number of observed paths representing the complete system operation may grow exponentially with the number of tasks. In addition, by using black-box identification methods, it is possible that the sequential order that the tasks must be performed is not represented in the model, reducing the fault detection capability or delaying the fault detection. In this paper, a two-level hierarchical approach for DES identification is proposed. In the higher level of the model hierarchy, the system is described by using some basic knowledge of its functioning provided by an expert, and in the lower level of the hierarchy, the behavior is described by black-box identified models for the system tasks. The modeling framework proposed in this paper reduces the number of observed paths needed for system identification and increases the fault detection capability. A practical example is used to illustrate the results of the paper.
Fichier principal
Vignette du fichier
2020_WODES_Marcos.pdf (847.71 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02558086 , version 1 (29-04-2020)

Identifiants

  • HAL Id : hal-02558086 , version 1

Citer

Ryan P C de Souza, Marcos V Moreira, Jean-Jacques Lesage. A hierarchical approach for discrete-event model identification incorporating expert knowledge. 15th Workshop on Discrete Event Systems, (WODES’20), Nov 2020, Rio de Janeiro, Brazil. pp. 275-281. ⟨hal-02558086⟩
54 Consultations
39 Téléchargements

Partager

Gmail Facebook X LinkedIn More