Weibull分布定时截尾试验数据情形下的数据填充算法  

The Quantile-Filling Algorithm for Weibull Distribution Under Type I Censored Data

作　　者：姜宁宁[1] 

机构地区：[1]中国科学院数学与系统科学研究院,北京100080

出　　处：《应用数学学报》2007年第6期986-995,共10页Acta Mathematicae Applicatae Sinica

基　　金：中国科学院国防科技创新基金(CXJJ-128)的资助项目.

摘　　要：利用处理非随机删失数据的矩不变准则及相应算法,本文针对Weibuu分布相同定时截尾型试验数据,通过控制算法中填充数据的分位概率,证明了通过迭代算法所得到的参数估计的相合性.进一步,通过对参数空间进行压缩,使得改进后的收敛算法所得到的参数估计同时具有相合性及不变性.作为本文算法的一个应用,我们还研究了利用填充后的虚拟完全样本构造可靠度的区间估计.最后,通过模拟计算验证了该算法具有较好的计算稳定性和可操作性.Based on the moment invariance criterion and corresponding algorithm which is proposed to deal with non-random censoring data, in this paper, the consistency of estimator is proved in the case of the Weibull distribution under type I censoring data, through controlling the quartile probability of the data filling algorithm. Furthermore, we modify algorithm through compressing the parameter space so that the consistency and invariability of the estimator can be simultaneously guaranteed. As an apllication, we use the virtual complete data obtained from the filling algorithm to get interval estemator of reliability. At the end, the simulation study shows that the algorithm has good stability and maneuverability.

关 键 词：矩不变准则 非随机删失 不变性 相合性 区间估计 

分 类 号：O211.9[理学—概率论与数理统计]