可靠性模型自助参数估计法  

Bootstrap method of parameter estimation in reliability model

作　　者：王智明[1] 杨建国[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200240

出　　处：《哈尔滨工业大学学报》2010年第5期820-823,共4页Journal of Harbin Institute of Technology

基　　金：高等学校全国优秀博士学位论文作者专项资金资助项目(200131)

摘　　要：为了提高可靠性模型参数估计精度,在分析传统可靠性模型的基础上,指出传统的最小二乘法估计和逐次线性概似估计因受随机变量假设分布的限制存在一定的局限性.用自助法给出了可靠性模型的参数估计,包括点估计和区间估计.计算结果表明:自助法所得参数的估计区间、标准差及模型标准残差均小于传统回归法所得结果.在可靠性模型参数估计中,自助参数估计法通过增加模拟次数可快速逼近参数真值,方法快速、有效,可提高模型精度.In order to improve the precision of parameter estimation of reliability model,based on the analyses of traditional reliability model,the limitations of least square method and gradual linear approximation because of their hypothetical distribution to random variable are pointed out,and the parameter estimators of point and interval are given by bootstrap method.The calculating results show that the estimated interval,standard error of parameter and the standard residual errors of the model are both less than those estimated by traditional method.Bootstrap method can approximate the actual value quickly than traditional way when increasing simulative times,so this method is feasible and effective and has a better accuracy than the latter.

关 键 词：可靠性模型 自助参数估计 最小二乘法 逐次线性概似 

分 类 号：TB114[理学—概率论与数理统计]