Abstract:

The reliability qualification experimental design about failure rate for exponential products was made based on Bayesian  decision theory. Considering the relation between the actual economic loss and the failure rate, the decision loss function was constructed so as to quantify loss. According to the posterior risk criteria, the loss caused by producer's risk and  consumer's risk were constrained. The maximum sample size and the failure number were determined in order to get the maximum expected net gain from sampling in type Ⅱ censoring way. The qualified determination time according with the product identification was deduced by posterior loss criteria. In all, the consumer can obtain the more rational and effective qualification scheme than only considering the decision risk. At last, an example was given to illustrate the validity and practicability.

Key words: expected net gain from sampling, reliability qualification experimental plan, Bayesian decision, exponential product

摘要：

利用Bayes决策理论，对指数型产品的失效率进行了可靠性鉴定试验方案的设计。根据实际经济损失与失效率之间的关系确定了决策损失函数,使损失真正量化。采取定数截尾方式，以验后风险准则下的弃真和存伪两类风险的约束为前提,依据抽样净益最大的原则确定了最大样本量和失效数。根据后验损失准则，得到满足产品鉴定的合格判定时间。为此，使用者能够从经济方面考虑获得比单纯用概率风险方案更加合理有效的鉴定方案。最后用算例验证了方法的有效性和可应用性。

关键词: 抽样净益, 可靠性鉴定试验方案, Bayes决策, 指数型产品