对数正态分布的最好线性无偏估计的快速计算  

A Fast Calculation Method of Best Linear Unbiased Estimation for Lognormal Distribution Product

作　　者：金星[1] 洪延姬[1] 王广宇[1] 叶继飞[1] 

机构地区：[1]解放军装备指挥技术学院,北京101416

出　　处：《机电产品开发与创新》2009年第6期8-9,共2页Development & Innovation of Machinery & Electrical Products

摘　　要：对数正态分布的最好线性无偏估计方法是工程中常用方法,但是目前只能用于样本量不超过20个的情况。通过将无穷区间内积分,变换为有限区间内积分,以及采用高效的自适应高斯积分方法,解决了次序统计量的均值和方差的快速计算问题,从而将对数正态分布最好线性无偏估计方法,推广到样本量为40个以上情况,为可靠性数据分析,提供了高精度的计算方法。The parameter estimation method of best linear unbiased estimation is a very useful method in engineering, which sample size is less than 20. Based on the integral function is transformed from infinite interval to limited interval, and the adaptive Gauss integral method is introduced, a fast calculation method is presented for the mean value and variance of order .statistics. Furthermore an optimized calculation method of best linear unbiased estimation is proposed for lognormal distribution product. The method can be used in sample size more than 40, which a high precision calculation method is offered for reliability data analysis.

关 键 词：对数正态分布 定数结尾试验 最好线性无偏估计 

分 类 号：O29[理学—应用数学] TB11[理学—数学]