长寿命产品在小子样缺失数据下的Bayes可靠性增长分析

在长寿命产品的可靠性增长试验过程中,由于人员、观测设备或其他方面的原因,可能会造成某些试验数据丢失或未观测到的现象。对这类小子样变总体缺失数据情形,提出了Bayes可靠性增长分析方法。首先利用Box-Tiao技术构造先验分布,然后利用非齐次Poisson过程原理和缺失数据的产生机制,得到可靠性增长缺失数据的似然函数,再用Bayes统计推断方法得到产品各研制阶段结束时的可靠性水平,同时给出了缺失数据下增长模型的拟合优度检验方法。最后通过一个示例说明了该方法在工程上的应用。

Bayes Reliability Growth Analysis for Long Life Products with Small Sample Missing Data

During the course of reliability growth tests for long life products, for the reason of people, equipment etc, some test data may be lost or can not be observed. A Bayes analysis method be proposed in order to solve the problem of reliability growth analysis with the condition of small sample missing data. Firstly, prior distribution is constructed based on Box-Tiao method. Then, the likelihood function is arrived according to non-homogeneous Poisson process and missing data occurring mechanism. Based on Bayes statistical inducing method, the reliability of product in different development periods can be arrived. Also, the fitting degree verified method for reliability growth model with missing data is given. At last, an example is given to illustrate the effective of the proposed method.