退化数据分析的EM算法(英文)  

EM algorithm for degradation data analysis

作　　者：徐安察[1] 汤银才[1] 

机构地区：[1]华东师范大学金融与统计学院,上海200241

出　　处：《华东师范大学学报（自然科学版）》2010年第5期38-48,共11页Journal of East China Normal University(Natural Science)

基　　金：国家自然科学基金(10571057,90718013);华东师范大学2010年优秀博士培养基金(2010049)

摘　　要：提出了用EM算法对产品可靠性进行分析.在退化模型中,当随机效应服从指数族分布时,推导了参数估计的一般公式并且通过模拟评价了随机效应分布选取的敏感性.最后分析了两个实际例子.Statistical inference procedures were proposed in assessing product reliability based on EM algorithm. The general formulae to estimate parameters were derived in the degradation model when its random effects follow an exponential family, and sensitivity of the choice of the distributions of the random effects was assessed via simulation. Two real data sets were analyzed.

关 键 词：退化数据 随机效应 EM算法 

分 类 号：O213.2[理学—概率论与数理统计]