|
Abstract:
|
The first step in the study of the reliability of a system is the estimation of the reliability of each component of the system. In some situations, however, the component that causes the system failure is not identified and we can only establish that it belongs to a set of candidate components to cause system failure. Cases like this are known as examples of failures with masked causes. For parallel and series systems, there is a high number of solutions in the reliability literature. From the best of our knowledge, we were not able to find works that deal with component's reliability in masked data scenario for more complex structures. In this sense, a Bayesian three-parameter Weibull model for component's reliability in any coherent masked system is proposed. The restriction of components failure times identically distributed is not imposed. The prior distribution was chosen subjectively and Metropolis within Gibbs algorithm is considered to simulate from posterior distribution. The proposed model presented great performance evaluated through several simulation studies and its practical relevance was demonstrated in a real data set of computer hard-drives monitoring.
|
