Contribution à l’analyse bayésienne en fiabilité et maintenabilité des systèmes mécaniques

Abstract

The Bayesian approach is a statistical method; it allows trend studies on the behavior of materials between periods or after a break in the life of these materials. It naturally includes consideration of information partially uncertain for support in modeling problem. The method is therefore particularly suited to the analysis of systems having mechanical and electronic complexity. The Bayesian techniques are used to reduce the size of estimation trials, improve the estimation of the parameters of product reliability by integrating the past (data available on the product concerned) and treat the case "zero" failure observed difficult to treat with conventional statistical approach. The work is based on the analysis and the choice of a prior distribution system or entity to define, without testing accelerated or aggravated behavior posteriori by calculating the probability density. The reliability tests are very expensive and very long especially if the components have a high reliability. Modeling considered the most practical distributions for the Bayesian (discrete or continuous) approach by applying virtual testing and the integration of expert opinions with a certain level of confidence.