恒加试验中几种线性无偏估计及其比较

A Few Linear Unbias Estimates and Their Comparison in Accelerated Life Test

  • 摘要: 在恒加试验中加速方程的二个系数ab的估计最为重要,常用的加权最小二乘估计\widehata与\widehatb在线性无偏估计类中并不是方差最小的。本文指出,这是由于忽视某些中间估计的恒相关性而使方差增大,考虑这些相关性之后,可以给出ab的BLUE的简化算法,本文还对ab设计一种新的加权最小二乘估计,其方差接近最小,而计算较为简便,最后,还给出二种加权最小二乘估计分别与BlUE的相对效率的下界,一个数值例子说明上述各种比较的效果。

     

    Abstract: It is important to estimate two coefficients a and b of accelerated equation in accelerated life test, and such a and b are often estimated by means of weight least square estimator (WLSE). Its algorithm is simple, but it cannot get a minimum variance because of ignoring the correlation between estimations of a and b. In this paper we designed a new quasi-BLUE for a and b. It can not only make the variance approach minimum, but also make the algorithm simple as well. Finally, this paper gave lower bounds of relative efficiency of WLSE and quasi-BLUE.

     

/

返回文章
返回