非正态分布时最小子样容量的确定

Determination of The Minimum Sample Size for Non-Gaussian Distribution

对可靠性试验中非正态分布前提下如何确定最少试验个数的问题进行了研究。提出了采用当量正态化方法 ,并综合考虑假设检验的两类错误 ,利用统计推断方法 ,在一定精度指标和置信水平下 ,给出估计母体均值和母体下限值时所需最少试件个数。

The number of specimen in reliability test is discussed for non-Gaussian distribution. The equivalent normalization method is given when the design variable does not come from Gaussian distribution. Considering the two types of errors in hypothesis testing, the minimum number of specimen required to estimate the mean and the percentiles of a random variable is obtained under a certain confidence level and a certain precision index. The result of the example indicates that the given method is practical.