Improved maximum likelihood estimation for component reliabilities with Miyakawa—Usher—Hodgson—Guess' estimators under censored search for the cause of failure

Abstract

A technique to improve, in life-tests for multi-component systems, the accuracy of the estimation of the failure rates in presence of a certain type of incomplete data (‘masked’ data) is discussed. A sufficient condition for some estimators studied by Miyakawa, Usher, Hodgson and Guess to be the maximum likelihood estimators is given and it is shown that such a condition is satisfied in the framework which we consider.