Vous êtes ici : Accueil » Kiosque » Annonce


Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?


8 juillet 2019

Predictive maintenance models for cyber-physical systems (CPS) with the consideration of sensor degradation

Catégorie : Doctorant

Université Technologie de Troyes, LM2S

Titre du sujet :

Modèles de Maintenance Prévisionnelle pour les systèmes Cyber-physiques (CPS) avec la prise en compte des impacts du processus de dégradation de Capteurs

Acronyme : MP2C

Mots-clés : Modélisation stochastique, maintenance prévisionnelle ; systèmes cyber-physiques ; pronostic de la durée de vie résiduelle ; dégradation de capteurs.

L’objectif principal de cette thèse est d'améliorer la performance de la maintenance prévisionnelle d’un système cyber-physiqueen développant :

1)Des modèles stochastiques pour la dégradation de capteurs et d’un réseau de capteurs prenant en compte les dépendances entre les capteurs ;

2)Des indicateurs et des méthodes pour quantifier l’impact du processus de dégradation des capteurs sur la performance de pronostic et de la maintenance prévisionnelle ;

3)Des modèles avancés pour la maintenance des systèmes cyber-physique, par exemple, modèle de l’optimisation conjointe de la maintenance prévisionnelle de la machine et de la maintenance de son réseau de capteurs.


Nowadays, industry 4.0 is a promising technological solution to improve the productivity and competitiveness of companies by creating more flexible and efficient organizations to produce higher quality products at reduced costs. From a maintenance point of view, intelligent production lines are composed of several Cyber-Physical Systems (CPS) that exchange information in real time via advanced connectivity (IoT: Internet of Things). A CPS can be an integration of a machine with its sensors, computers and actuators to allow the CPS to be able to automatically measure the degradation level, to predict the evolution of degradation, to calculate the Remaining Useful Life (RUL), and to perform certain maintenance actions on the machine. The success of this predictive maintenance process depends on several factors, especially the proper functioning of the CPS’s sensors. However, these sensors do not always work perfectly due to their deterioration over time. Given important impacts of the sensor degradation on the RUL estimation, as well as on the performance of the predictive maintenance, it has been rarely addressed in the literature. For this reason, the main objective of the project is to improve the performance of predictive maintenance by taking into account not only the machine degradation, but also that of its sensors. In accordance with the previous objectives, the project can be structured into the three following phases:

The degradation and maintenance models developed in the project can help to increase the accuracy of the RUL prediction and to improve the quality of the maintenance decision-making, the failure prevention and the logistics support. In this context, not only the maintenance costs and logistics support costs, which play a very important role in the total operating cost of companies, but also the performance and the RAMS (Reliability-Availability-Maintainability - Safety) of the machine, are also improved. This work can also contribute to an evolution of maintenance practices in industry 4.0 context.

Candidate profile: Master or engineer with a dominant focus on statistics or signal processing. The implementation of the project requires knowledge in probabilistic and statistical approaches such as stochastic modelling, probability analysis, statistical estimation and Monte Carlo simulation. Knowledge about the intelligent sensors would be highly appreciated.

Contact: mitra.fouladirad@utt.fr


Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2019.