稳健统计过程控制图的构建与比较

Construction and Comparison of Robust Statistical Process Control Charts

在构建标准差过程控制图时,控制图的控制上下限十分不稳健,极易受到异常情况的影响,导致控制过程可能发生错报或者漏报。文章对4种经典稳健标准差估计方法进行比较,之后提出了一种基于样本筛选的稳健标准差控制图,考虑了在阶段I过程中常见的4种异常值类型,以均方误差值作为稳健估计效果优劣的评判准则,结果表明新提出的基于样本筛选的稳健估计方法在大部分污染情况下都有着非常好的稳健估计效果。

During the construction of standard deviation process control chart,the control upper limit and lower limit of the control chart are not robust and easily affected by abnormal conditions,which may lead to misstatement or omission of statement in the control process.This paper compares four classical robust standard deviation estimation methods,and then proposes a robust standard deviation control chart based on sample screening,withfour types of outliers commonly seen in Stage I process considered,and the mean square error value used as the evaluation criterion for the robustness estimation effect.The results show that the new robust estimation method based on sample screening has a very good robust estimation effect in most pollution cases.