双参数指数分布异常数据的检验  

Test of Outliers in Two-Parameter Exponential Distribution

作　　者：梁米 李云飞[1] LIANG Mi;LI Yunfei(College of Mathematics and Information,China West Normol University,Nanchong Sichuan 637009,China)

机构地区：[1]西华师范大学数学与信息学院,四川南充637009

出　　处：《西华师范大学学报（自然科学版）》2021年第4期355-360,共6页Journal of China West Normal University(Natural Sciences)

基　　金：西华师范大学英才科研基金项目(17YC381)。

摘　　要：针对样本服从双参数指数分布场合下的异常大和异常小数据,提出了一种新的检验方法。首先利用样本数据中异常数据的离群性和平均寿命点估计的跳跃度确定可能的正常数据在次序统计量中的位置,并由此出发向左、向右逐步推移,然后基于双边截尾数据场合下参数σ的最好线性无偏估计,构造新的检验统计量,并给出其精确分布,确定判定标准。最后通过多次重复模拟试验说明所给出的方法具有可行性。This paper proposes a new test method for large and small abnormal data in two parameter exponential distribution.Firstly,the position of possible normal data in order statistics is determined by outliers of abnormal data in sample data and the jump degree of the average life point estimation,from which it is set out step by step to the left and right.Then,a new test statistics is constructed on the basis of the best linear unbiased estimation of parameter in the case of bilateral censored data.In addition,its exact distribution is given and the criterion is determined.Finally,the feasibility of the proposed method is illustrated by repeated simulation experiments.

关 键 词：双参数指数分布 最好线性无偏估计 异常数据 检验统计量 

分 类 号：O212.1[理学—概率论与数理统计]