论文部分内容阅读
一、引言作者在[1]中曾对集成电路(TTL中速“与或非”门及“与非”门,型号B_3)高温贮存寿命试验的失效数据按照简单线性无偏估计(GLUE)方法进行处理,而采用GLUE方法的假设之一是失效数据服从威布尔分布,这就产生了这些失效数据是否服从威布尔分布的问题。本文介绍四种常用的分布假设检验方法,并用柯尔莫洛夫—斯米尔诺夫(Kolmogorov-Smirnov)检验法(简称K-S法)和置信区间检验法(d检验)对四种温度的失效数据进行检验。同时,用Bartlett检验法对形状参数m相等的假设进行检验,还对m估计量的置信区间
I. INTRODUCTION In [1], the failure data of the high-temperature storage life test of integrated circuits (TTL medium speed NAND gates and NAND gate and type B_3) were evaluated according to the simple linear unbiased estimation (GLUE) method , And one of the assumptions used in the GLUE approach is that the failure data obey the Weibull distribution, which raises the question of whether these failure data obey the Weibull distribution. In this paper, four commonly used distributional hypothesis testing methods are introduced, and four kinds of temperature failure data (Kolmogorov-Smirnov test and KS test) To test. At the same time, Bartlett test is used to test the hypothesis that the shape parameter m is equal, and the confidence interval of m estimator