基于混合Beta分布的成败型产品Bayes可靠性鉴定试验方案研究

A Bayes Plan of Reliability Qualification Test Based on the Mixed Beta Distribution for Success/Failure Product

研究了基于混合Beta分布的成败型产品可靠性的Bayes鉴定试验方案。通过引入继承因子,合理地考虑了产品在设计和改进过程中的各种信息,并把继承因子看作随机变量,使用混合Beta验前分布建立了可靠性鉴定方案的Bayes决策模型。算例结果表明,在小样本情况下,经典鉴定方案得到的试验样本量过于保守;传统Bayes方法虽利用了产品的先验信息,但却忽略了它们之间的差异;而本文的方法较好地解决了上述问题,并具有方法上的合理性和工程上的实用性。

A Bayes plan of reliability qualification test for success/failure product was studied based on the mixed Beta distribution. The Bayesian decision model of the plan was established based on the mixed Beta distribution, which is made by introducing the factor of inheritance regarded as a random variable into Bata distribution to make the plan accord with engineering practice. The outcome of the example given in this article indicates that the quantity of the test are too conservative according to classical methods using small binomial samples; although traditional Bayesian analysis can consider test information from related or similar products, it ignores the differences between such products; the method proposed in the paper can solve the problem successfully, rationally and practically.