Nonparametric Analysis on System Availability: Confidence Bound and Power Function

Abstract

In this study we consider the steady-state availability, denoted A, of a system with distribution-free failure and repair time. In particular, we are interested in constructing a lower confidence bound and a testing framework for A. We first show that the natural estimator  of A, defined as the ratio of the failure time sample mean to the sum of the failure time sample mean and the repair time sample mean, is strongly consistent and asymptotically normal. Then using the asymptotic distribution of Â, we develop a lower confidence bound and a hypothesis test for A. Finally, a numerical simulation study is conducted in order to illustrate the performance of  in applied inference about A.