系统可靠度置信下限估计方法及其应用

针对寿命服从指数分布的随机截尾样本，研究单元及系统可靠度的评估方法，给出了系统可靠度置信下限估计方法，该方法同样适用于工程中常见的定时截尾样本与定数截尾样本的系统可靠度评估问题。首先通过分位数填充算法将随机截尾数据补充成虚拟完全样本，然后基于信仰推断，给出系统内每个单元失效率的信仰分布，最后根据指数分布的特点，从失效率角度出发，采用蒙特卡罗法对单元失效率进行随机抽样，结合系统可靠性模型得到系统失效率的分布，在给定置信水平下即可得到系统失效率上限，进而得到工程上特别关注的系统可靠度置信下限。

Method for system’s reliability confidence lower limit estimation and its application

Based on random censoring data whose life follows exponential distribution, unit and system reliability evaluation methods are studied, and a method of estimating the reliability lower limit is proposed in the case of the exponential distribution under random censoring data, which is also applicable to type Ⅰ censoring data and type Ⅱ censoring data. Firstly, the censoring data are complemented into virtual complete data by use of the quantile-filling algorithm. Based on the fiducial inference, the failure rate distribution of each unit is given. According to the characteristics of the exponential distribution, in view of the failure rate, unit failure rate samples are randomly extracted by Monte Carlo method. In addition to system reliability model, the distribution of system failure rate is obtained, therefore the upper limit of the system failure rate at the given confidence level, and the lower confidence limit of the system reliability are obtained.