Confidence Interval Estimation of the Reliability of Multicomponent Systems Using Component Test Data

摘要

A method is developed for the construction of confidence intervals for a general class of functions of binomial parameters called system reliability functions. These system reliability functions express the reliability of a multicomponent system as a function of the reliabilities of the components of the system. The method derived uses the exact multivariate binomial distribution of component test data. The problem is formulated as a general mathematical programming problem and numerical methods are developed for the solution of the problem. The method is applied to several specific systems and detailed computational procedures are developed for each. The method derived is compared with other currently available methods for constructing confidence intervals for system reliability based on component test data and the results of this comparison indicates the possibility of obtaining and appreciable reduction in the average length of confidence intervals for a given sample size and confidence coefficient by the use of the method of this report. The use of these confidence intervals for the construction of statistical tests of hypotheses concerning system reliability is also discussed. (Author)