An ad hoc Bayesian method for determining lower confidence limits.

Abstract

This thesis proposes and analyzes an ad hoc Bayesian method for determining lower confidence limits for estimates of mission reliability of single components using pass/fail test data. Several test sequences with different underlying reliabilities are simulated. After each trial the lower confidence limits are calculated using the Classical method, the Bayesian method and the proposed method. This is followed by an accuracy simulation to estimate the accuracy of the proposed method. The proposed method shows promise for future usefulness in that it is capable of demonstrating a high degree of reliability with a relatively small number of trials while maintaining some degree of protection against accepting unreliable components.