Evaluating the reliability of repairable systems

Abstract

The Weibull-Poisson process (WPP) for representing the reliability of complex repairable systems is discussed. The emphasis is on estimation and other statistical methods for this model when data have been generated by multiple systems. Examples and procedures specifically illustrating these methods are given for several real-world situations. In addition to maximum likelihood estimation methods, goodness-of-fit tests and confidence interval procedures are discussed and illustrated by numerical examples. It is noted that in the case of one system the model reduces to a model for reliability growth. Confidence intervals for the WPP shape parameter and growth rate are given.<>