多元预控图监控准则优化研究

Optimization of monitoring rules for multivariate pre-control charts

多元预控图将一元预控图的监控变量由一元扩展到多元,是分析和监控两个及两个以上相关质量特性的有效方法,是复杂生产过程中控制产品质量的重要工具,但也存在精度偏低和监控效率低下等问题。为此,在一元监控图基础上,为降低多元预控图的误判概率,按增加单个样本所能提高的概率来调整抽样数,由5个增加到10个,在新的抽样数条件下调整监控准则,由4个增加到6个,并采用平均链长度(ARL)为基准,对新旧监控准则的监控效果进行比较并获取最优监控准则,提升了多元控制图的的监控精度和效率。

Multivariate pre-control chart extends monitoring variable from one variant to multiple, which is an effective method to analyze and monitor two or more quality characteristics ,and an important tool to control product quality of production process as well.However,problems are available such as low precision and low monitoring efficiency.On the basis of pre-control chart, in order to reduce the risk of false-judgment, sampling amount has been increased from 5 to 10 in accordance with probability iucreased by every additional sample,and monitoring rules has been adjusted from 4 to 6 based on new sampling amount.With average run length （AVL）as standard,the effect of new and old monitoring rules has been compared and then the best one could be obtained,which has promoted the monitoring precision and effi- ciency of Multivariate pre-control chart.