Although reliability theory and analysis is a well-established discipline in engineering applications, many analysts often find it challenging to conduct reliability analysis in practice. Seldom the theoretical models and methods are directly applicable. There are usually a number of ways of approaching a reliability problem. The reliability theory offers a number of possible advanced models and techniques, but what is the proper approach in a specific case is not easily extracted from the theory. The general impression is that a probability model needs to be specified, and parameters estimated. The focus of the analysis is then the parameters, quantities that are often difficult to interpret. In this paper we point at the need for a rethinking of this tradition. An approach where focus is observable quantities would make the analysis easier to understand, and the message from the analysis is more direct. In the paper we present and discuss such an approach, within a predictive Bayesian framework.
展开▼