融合先验加速退化与外场退化信息的可靠性评估方法

针对性能退化过程服从Wiener过程的产品,运用贝叶斯统计推断法,提出了一种融合同类产品加速退化试验(accelerated degradation test, ADT)信息与个体外场退化信息的可靠性评估方法。考虑到内、外场应力环境之间的差异,提出了基于修正系数的Wiener过程双参数修正模型;分别建立了步进ADT数据模型与外场退化数据模型,得到各应力下分布参数估计值;利用加速因子,将各加速应力下分布参数估计值折算到正常工作应力下,构成未知参数先验分布的数据样本;采用CvM检验法,确定未知参数最优先验分布类型及其超参数估计值;构建了外场退化数据下未知参数的后验分布函数,采用马尔可夫蒙特卡罗方法,得到参数的后验分布均值。通过实例分析验证了所提方法的正确性和实用性,结果表明,本方法可有效处理先验ADT信息与外场退化信息之间的融合评估问题。

Reliability assessment method with integrated prior accelerated degradation and field degradation data

Aiming at the product which its performance degradation process obeys to the Wiener process, a reliability assessment method for accelerated degradation test (ADT) data of like products and field degradation data of individual is put forward by using Bayesian statistical inference. Considering the difference between field stress environment and laboratory stress environment, the Wiener process double-parameters modified model with the correction factors is constructed. The step-stress ADT data model and the field degradation model are built to obtain estimated values of distribution parameters under each stress. The accelerated factor is used to convert these estimated values under accelerated stress into regular stress and constitute data sample for prior distribution of unknown parameters. The CvM checkout method is used to determine the optimal prior distribution of unknown parameters and its estimated values of hyper parameters. Posterior distribution function of unknown parameters under field degradation data is built and the markov chain monte carlo (MCMC) method is used to obtain mean values of the posterior distribution. Accuracy and practical applicability of the present method is verified by an example. Result shows this method can deal with the integrated assessment problem for prior ADT data and field degradation data.