基于Copula函数选择的多元退化产品可靠性模型

Reliability Model of Multivariate Degradation Product Based on Copula Function Selection

针对含有多个退化量的产品可靠性评估问题,提出了一种基于Copula函数选择的可靠性建模方法。首先,采用Anderson-Darling统计量对退化量进行拟合优度检验,选择其最优分布函数;其次,绘制退化量分布函数中的分布参数随测量时间的变化曲线,判断其与时间和应力的相关性;再次,给出一种选择Copula函数的综合方法,该方法结合了三种常用的选择方法;同时,采用两步似然估计法对未知参数进行了评估;最后,通过硫化丁腈橡胶O型密封圈进行恒定应力加速退化试验对方法的正确性进行了实例验证,并且与二维正态分布模型、形状参数为常数时的可靠性评估结果进行了对比分析。分析结果表明,给出的方法具有更好的实用性和更广泛的应用。

According to the reliability assessment of products with multiple degradation,a method for build reliability modeling based on Copula function select was presented. First of all,the goodness of fit was tested by Anderson-Darling statistic,and the optimal degradation distribution function was selected. Secondly,the correlation of distribution parameters with time and stress was determined by the curve of distribution parameter with measurement time. Then,an integrated method for selecting Copula function was presented,which combines three common selection methods. And two step likelihood estimation method was used to evaluate the unknown parameters. Finally,seal through vulcanization of nitrile rubber O type constant stress accelerated test method for the degradation of correctness was verified,and two dimensional normal distribution shape parameter model,the reliability evaluation for constant were compared. The results show that the method presented has better practicability and wider application.