三种统计方法在润滑脂性能验证分析中的应用

The Application of Three Statistical Methods in Performances Verification Analysis of Grease

用四分位稳健统计法,迭代稳健统计法和经典统计法(用格拉布斯准则剔除离群数据)对38家实验室测定的极压复合锂基润滑脂的工作锥入度,延长工作锥入度,滴点,锥网分油及磨斑直径的数据群进行了分析,以评估实验室的检测能力。对于总体服从正态分布或接近正态分布的数据群,三种统计方法的结果基本吻合,但建议使用四分位稳健统计法或迭代稳健统计法进行分析,若存在离群数据,应采用经典统计法进行分析。对于明显偏离正态分布,但不存在离群数据的数据群,建议使用迭代稳健统计法进行分析,若存在离群数据,则采用经典统计法进行分析。

In order to evaluate the testing ability of the laboratory,the data group of the worked cone penetration,prolonged worked cone penetration,dropping point,oil separation(conical sieve method)and wear spot diameter form 38 laboratories were analyzed by using the quartile robust statistical method,iterative robust statistical method and classical statistical method(excluding outlier data using Grubbs Criterion).For the data group with normal distribution or close to normal distribution,the results of the three statistical methods were basically consistent.But it was recommended to use the quartile robust statistical method or iterative robust statistical method for analysis.If there was outlier data,the classical statistical method should be adpoted for analysis.For data groups that deviatd significantly from normal distribution,but did not have outliers,iterative robust statistical method was recommended for analysis.If there were outliers,classical statistical method was preferred for analysis.