威布尔分布场合无失效数据的失效概率估计方法  

THE ESTIMATION TO FAILURE PROBABILITY FOR ZERO-FAILURE DATA IN WEIBULL DISTRIBUTION

作　　者：贾祥[1] 蒋平[1] 郭波[1] 

机构地区：[1]国防科技大学信息系统与管理学院,长沙410073

出　　处：《机械强度》2015年第2期288-294,共7页Journal of Mechanical Strength

基　　金：国家自然科学基金项目(71371182)资助~~

摘　　要：随着产品可靠性的提高和成本的增加,以及研制周期的限制,现在可靠性试验常采取小样本、短周期,因此试验中收集到的往往是无失效数据,配分布曲线方法在处理这类无失效问题中应用较广。方法的关键步骤就是失效概率的估计,但是针对威布尔分布场合的失效概率估计,研究还比较缺乏。根据形状参数的不同取值范围,利用分布函数的特点和凹凸性,来确定对应的失效概率的取值范围。在实际操作中,可通过工程实践或者专家信息确定形状参数的取值范围,并进一步采用Bayes方法得到失效概率的估计值。仿真算例和轴承实例的计算分析也证明了该方法的有效性和稳健性。Nowadays, reliability tests are often carried out in small sample size and short duration, due to the improvement of product reliability and increasing of cost, as well as the restriction of development time. Therefore, zero-failure data are often collected in such tests. The Match Distribution Curve method is widely applied to handle zero-failure data. The key step to exercise the method is to estimate the failure probability, which is lack to research in the case of Weibull distribution. To estimate failure probability in Weibull distribution, the intervals of failure probability were proposed in this paper using the concavity or convexity and property of distribution function according to the intervals of shape parameter in Weibull distribution. Furthermore, to practically use the proposed method, the approximate value of shape parameter could be determined by engineering experience or expert information and the estimation of failure probability was calculated by Bayes method. A simulation case and a real example are presented to prove that this method is effective and robust.

关 键 词：无失效数据 失效概率 BAYES估计 

分 类 号：TB114.3[理学—概率论与数理统计]