Abstract: This paper introduced the statistical methods which are used to filter out abnormal value, including the Pau Ta＇s Criterion(3σcriteria), the Grubbs’ Criterion, the Chauvenet’s Criterion and the Dixon＇s Criterion. Furthermore, their feasibility and effectiveness in processing gross errors of safety monitoring data were discussed here.

摘要： 通过引入统计学中用于滤除异常值的拉依达准则（3σ准则）、格拉布斯准则、肖维勒准则、狄克逊准则，讨论了利用这些判别准则进行安全监测数据粗差处理的可行性和有效性。