Imperfect Maintenance: Stochastic Modelling, Statistical Inference, Software Development and Real Case Study

français

Speciality : Mathématiques Appliquées

16/03/2018 - 09:45 Mr Laurent Doyen (Université Grenoble Alpes) Auditorium - RDC - Batiment IMAG

Keywords :
  • Probabilistic modelling
  • Recurrent event
  • Stochastic Process
  • Reliability
  • Repairable system
Maintenance is carried out on repairable industrial systems throughout their life cycle to keep them in, or restore them to, an operating state. Maintenance, by providing an essential contribution to the operational system reliability, plays a great part in risk management and constitutes a crucial element in the performance of an industrial installation. The basic assumptions on maintenance effectiveness are known as minimal maintenance or As Bad As Old effect and perfect maintenance or As Good As New effect. Reality falls between these two extreme cases. This is known as imperfect maintenance. 
This synthesis of my research activity is organized around a real case study corresponding to the successive failure and maintenance times of identical engines of Brazilian mining trucks. First, classical recurrent event models incorporating imperfect maintenance effects are presented. Statistical inference methods are then proposed. Both frequentist and Bayesian inference are considered. Special emphasis is given to semiparametric estimation. In fact, usual profile likelihood method does not lead to consistent estimator in this case. But we have proved that the consistency property can be recovered by smoothing the infinite dimensional parameter in the profile likelihood function. Particular frameworks have been developed in order to take into account the different maintenance types: corrective maintenance, time-base or condition based preventive maintenance. Finally, models with random unobserved maintenance effects are considered. EM types algorithms have to be used for inference in this case. Most of the proposed methods are implemented in our R package VAM.

Raporteurs:

  • Mr Edsel A. Peña (Professeur - University of South Carolina (USA) )
  • Mme Sophie Mercier (Professeur - Université de Pau et des pays de l'Adour )
  • Mr Henry Lindqvist Bo (professeur - NTNU Trondheim (Norvège) )

Examinators:

  • Adeline Leclercq-Samson (Professeur - Université Grenoble Alpes )
  • Mr Olivier Gaudoin (Professeur - Grenoble INP )
  • Mr Emmanuel Remy (Ingenieur R&D - EDF R&D )