Reliability Acceptance Sampling Plans Based upon Prior Distribution. Volume III. Implications and Determination of the Prior Distribution.

摘要

This report deals with the determination of prior density for the case when the parameter of the failure distribution is considered to be random. First, the implication of considering the parameter as a random variable on some aspects of reliability analysis are examined. The effect of the population heterogeneity model on acceptance sampling and the effect of individual heterogeneity model on failure distribution, acceptance sampling, burn-in and system reliability are considered. The determination of the functional form of the prior distribution and the reasons for choosing a conjugate prior are discussed. Methods are given for estimating the prior parameters based upon data from various sampling situations. A comprehensive analysis of some observed failure data, including a study of the efficiency of the estimators, is also presented. (Author)