Skip to Main content Skip to Navigation
Theses

Aide à l'optimisation de maintenance à partir de réseaux bayésiens et fiabilité dans un contexte doublement censuré

Franck Corset 1
1 IS2 - Statistical Inference for Industry and Health
Inria Grenoble - Rhône-Alpes, LBBE - Laboratoire de Biométrie et Biologie Evolutive - UMR 5558
Abstract : The first part deals with application of Bayesian networks in maintenance and proposes a methodology to build a network from expert judgment. In order to evaluate probabilities, networks is seen as a saturated log-linear model. Some tools are defined in order to identify the most critical variables. Some maintenance tasks are integrated as new nodes of the Bayesian Network. An integration of experience feedback data is proposed by a Bayesian inference, by quantifying confidence given to experts judgment. The second part deals with reliability with doubly censored data. We study the asymptotic properties of the maximum likelihood estimators for the law of Weibull. A Bayesian inference is also proposed with informative and non informative priors. Moreover, human factors are described by hidden variables, which represent lack of information during maintenance operations. We solve this problem by maximum likelihood and by a Bayesian inference.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00003889
Contributor : Franck Corset <>
Submitted on : Wednesday, December 3, 2003 - 4:37:59 PM
Last modification on : Monday, February 10, 2020 - 4:36:45 PM
Long-term archiving on: : Friday, April 2, 2010 - 6:58:16 PM

Identifiers

  • HAL Id : tel-00003889, version 1

Collections

Citation

Franck Corset. Aide à l'optimisation de maintenance à partir de réseaux bayésiens et fiabilité dans un contexte doublement censuré. Mathématiques [math]. Université Joseph-Fourier - Grenoble I, 2003. Français. ⟨tel-00003889⟩

Share

Metrics

Record views

1161

Files downloads

521