贝叶斯估计与极小子样可靠性试验的一种策略

本文将贝叶斯方法应用至系统的可靠性试验之中,提出了一种进行极小子样系统可靠性试验的策略.该策略可以充分地利用经验信息,达到在小子样系统可靠性试验时仍可获得合理之结果的目的,因此,具有较高的实用及经济价值.

Accuracy of Results of System Reliability Test with Minimum Sample Size Improved with Bayes' Method

Castly experiments make it necessary to limit reliability test to small sample size. As a result, the accuracy of test results is much below desirable level. By introducing Bayes' method, the authors believe that they succeed in improving the accuracy of the results of reliability test with minimum sampie size. Usually much prior information exists before reliability test begins. Bayes' method can make use of prior information. Such prior information includes: (1) life distributions of similar systems; (2) life distributions of factory-produced components; (3) life distributions of installed components obtained from records of operational failures of these components. In Bayes' method, prior information is essentially treated as supplementary experimental data. Thus the limitation of small sample size is overcome to a considerable extent and accuracy of test results is much improved. The authors give an example of a series system and show that the introduction of Bayes' method gives quite good reliability estimates. The authors believe that Bayes' method is equally applicable to parallel systems and series-and-parallel systems.