电子器件生存概率及置信度的Bayes计算方法

Bayesian method to determine the survival probability and confidence of electrical devices under certain radiation environments

针对工程实践中经常遇到的小子样问题,分别以无信息分布、正态分布、产生正态性能分布作为电子器件生存概率随机变量的验前分布,给出了二项抽样实验验后生存概率密度的具体形式,根据某电子器件电流损伤阈值实验数据,得到特定辐射环境下器件生存概率的Bayes点估计、一定置信度下的双边估计、单边估计。仿真结果表明,产生正态性能验前分布要优于无信息分布和正态验前分布,该方法可以应用到工程实践中。

Calculation method to determine the survival probability of electrical devices under certain radiation environments is presented.As for the commonly encountered small sampling problems in engineering practice,uniform distribution,normal distribution and normal generated distribution are assumed as the apriori survival probability of the electrical devices concerned,then the corresponding posterior distributions with respect to binomial experiment are given.Finally,based on the experimental damaging threshold data in the form of electrical current,the point and the interval estimations of the devices’survival probability with certain degree of confidence under designated radiation environments are calculated.Results show that normal generated distribution is superior to the other two apriori distribution,which is more applicable in engineering practice.