基于休哈特控制图和累积和控制图在多变异工序中的应用

Monitoring Process with Multiple Source of Variation Using Shewhart and Cusum

传统的休哈特控制图在多变异源过程控制中会出现虚发警报概率过高。文章提出了一种多变异源控制图方法解决虚发警报过高的问题。通过对多个变异源的方差分析,基于对部分方差的估计,正确估计了均值的标准差,重新界定了控制图的控制限,建立了多变异源控制图,使虚发警报概率回复到原来的水平。由于累积和控制图在检测过程均值的小幅度偏移时比休哈特控制图更为有效,而在偏移较大时,休哈特控制图效果更好。文章采用休哈特与累积和联合控制图监控多变异来源过程的均值,对于各变异源的标准差分量,应用比较简便的休哈特单值控制图控制,取得了较好的效果。

The high probability of false alarm happened when traditional shewhart control charts are used for process that exhibit multiple source of variation. The paper addresses a new control system for process with multiple sources of variation to reduce the probability of false alarm. Based on the analysis of variance and variance components estimate for each source of variation, the paper defined the control limits again and presents control methods for multiple sources of variation. Because Cumulative sum（Cusum） control chart is more effective for controlling small shift than traditional shewhart control chart and the later is good at large shift., In this article, the two charts combined to control the xbar and individual chart is used to control variance components.