GSCOP-RUB-GCSP

Soutenance de thèse de Asma Abu Samah (GCSP) le 08 novembre 2016 à 14h en amphi Gosse - Site Viallet Grenoble INP

Intitulée ; Event based probabilistic approach for proactive maintenance to improve production capacities in HMLV industries
Membres du jury :

  • M. François, PERES Professeur, Ecole Nationale d’Ingénieurs de Tarbes (ENIT), Rapporteur
  • M. Philippe WEBER Professeur, Université de Lorraine, Ecole Supérieure des Sciences et Technologies de L’Ingénieur de Nancy (ESSTIN), Rapporteur
  • M. Sha’ri MOHD YUSOF Professeur Ir., Université de Technologie Malaisie de Kuala Lumpur, Examinateur
  • M. Michel TOLLENAERE Professeur, Grenoble-INP, Examinateur
  • M. Stéphane HUBAC Manufacturing Science Senior Expert Advanced Process and Equipment Control, STMicroelectronics-Crolles, Invité
  • M. Eric ZAMAÏ Maître de conférence HDR, Grenoble-INP, Directeur de thèse
  • M. Muhammad Kashif SHAHZAD Docteur, Institut National des Sciences Appliquées (INSA) de Lyon, Co-Encadrant dethèse

Abstract :

This research is carried out in the context of high-mix low-volume (HMLV) production environment. It is a production environment characterized by increasing demand volumes with product diversity and reducing product life cycles. This is a highly competitive production environment which require frequent changes in equipment settings and process recipes for product diversity and continuous introduction of new technologies to cope with increasing demand for more functional products. Moreover, ramp up production becomes highly critical because of reducing product life cycles. Therefore, to be successful in this highly competitive environment, we need stabilized and optimized production capacities to cope up with these challenges.

In such context, our objective are set as:

1) Minimize corrective maintenance
2) Improve preventive maintenance

2 integrated approaches was developed and presented as follows,

The first approach focuses on stopping the production equipment as a result of unscheduled breakdown if and only if it is the source of breakdown. This approach also concludes that to avoid confusion in failures and causes diagnosis, module must be benchmarked instead of equipment. This approaches is based on BN and uses contextual data instead of sensors data.

The second approach focuses on the extraction of time bound failure signatures which guarantees to generate an alert failure prior to specified time such that proactive measures can be taken to avoid failure occurrences.

This work is a part of the European INTEGRATE project.