在质量过程控制中,不合格品率小漂移的检测会受到样本的误分类以及小样本量的影响而产生严重的质量误判。本文首先针对样本的分类误差建立误分类修正数据模型实现样本观测值误分类修正,再结合DEWMA p控制图对参数小漂移的敏感性和对样本量的宽容特性,开发了带有误分类修正的DEWMA p控制图(MisC-DEWMA p)。实验结果表明,在对不合格品率的检测,特别是基于小样本的相应检测中,MisC-DEWMA p控制图对不合格品率小而持久的漂移具有更高的敏感性和精确性。In the quality process control, the monitoring of potential small shift at nonconforming proportion will suffer from misclassification and the small-size of sampled observations and thus present kind of a misjudge. In this paper, a data-modified model is firstly established to correct the sample data carrying misclassification, and then DEWMA p Control Chart with Misclassification Correction (MisC-DEWMA p) is proposed, equipped with the strong small-shift-sensitivity and small-sample-kindness of DEWMA p control chart. The numerical and real data experiment results show that, for a process with small persistent shift of nonconforming proportion, MisC-DEWMA p control chart provides a higher sensitivity and accuracy.
传统Shewhart-p控制图只对单一属性的不合格品率进行监控,在过程发生偏移时有一定的滞后性。为提高不合格品率控制图的精度,提出一种多元指数加权移动平均不合格品率(multivariate exponentially weighted moving average p, MEWMA-p)控制图。该控制图将多个属性的不合格品率应用于多元指数加权移动平均控制图,可同时对多个属性进行监控,并且对于小范围的偏移更加敏感。对比分析同等偏移程度下指数加权移动平均不合格品率(exponentially weighted moving average p, EWMA-p)控制图与MEWMA-p控制图的平均运行长度(average run length,ARL)结果,并通过模拟仿真说明该方法的有效性。