Skip to Main content Skip to Navigation
Theses

Pronostic des événements de défaillance basé sur les réseaux de Petri Temporels labellisés

Abstract : The deployment of decision-support tools increases agility efficiency while reducing the cost of maintaining proper operation. Accidental or intentional shutdowns have disastrous and costly consequences. The scientific community of discrete event systems (DES), has been interested in the cause-effect relationships between certain nominal and dysfunctional states, to propose solutions responding to this problem. Our work is part of the framework of a steering of a system subject to failure events (FE). We have developed a model-based prognostic approach, which predicts the earliest date of occurrence of an FE, allowing the operator to plan repair interventions on components before altering the proper functioning. We modeled the system using labeled temporal Petri nets (RdPTL), since the analysis of its behavioral model is based on the sequentiality and date of occurrence of events. We have represented these dynamics through modeling in a context of mode analysis, limited to 3 modes of operation (nominal, degraded and critical). From its accessibility graph, we have built a prognosticator, which allows us to identify all the sequences of events ending with an FE. We used the notion of parameterization of the system states i.e. the introduction of a clock and a system of clock inequation (SE) for each state of the system. The states obtained from the discretization of time are then grouped in a single state and the SE will determine the values of the clocks. The prognosis cannot always be guaranteed. We then established the prognosticability property, to distinguish the sequences that are prognosticable from those that are not. To validate our approach, we chose a battery cell as a benchmark and the INA tool to generate the accessibility graph.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-03510389
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, January 4, 2022 - 2:46:41 PM
Last modification on : Sunday, June 26, 2022 - 3:30:37 AM

File

these.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03510389, version 1

Citation

Redouane Kanazy. Pronostic des événements de défaillance basé sur les réseaux de Petri Temporels labellisés. Automatique / Robotique. Université de Lyon, 2020. Français. ⟨NNT : 2020LYSEI132⟩. ⟨tel-03510389⟩

Share

Metrics

Record views

74

Files downloads

38