Uso de métodos bayesianos em testes de vida acelerados no controle da qualidade de produtos industriais

Abstract

In this work, we introduce a Bayesian approach for quality control of industrial products, assuming units in test under stress levels higher than the usual level. We assume di¤erent distributions for the lifetimes of the units under a type II censoring mechanism and a general stress-response model that includes the usual accelerated life test models. Using a predictive density of a future observation, we present some criteria to be used by engineers in the quality control of a batch of industrial products. These criteria permit a choise for the best time in test or for the best stress level in the quality control test. The quality control procedure is based on the observed proportion of failures tested during a specified time and a specified stress level. We use the software WinBugs (Spiegelhalter et al, 1999) to obtain the posterior summaries of interest. keywords: Accelerated lifetime test, Bayesian Inference, Markov Chain Monte Carlo, Quality Control.