1　理论方法本文研究剔除因过失误差而产生异常数据的方法。已知实验数据在理论上服从正态分布 ,希望检验该组测定值中是否混有异常值 ,若有 ,则予以剔除 ,使余下数据服从正态分布。本文提出综合采用χ2 检验法和Grubbs检验法逐步对收集到的数据进行筛选。将收集到的数据组 ,用 A theoretical method This article studies the method of eliminating abnormal data caused by fault errors. It is known that the experimental data are theoretically subject to normal distribution, and it is hoped to test whether the abnormal values ​​are mixed in the measured values ​​of the group. If so, the data are excluded and the remaining data are subject to the normal distribution. This paper proposes a comprehensive screening of the collected data using χ2 test and Grubbs test. Will be collected data set, use