讨论了采用EWMA图进行工序过程监控时控制图的优化设计问题。首先提出了用马尔可夫链的方法对EWMA图的平均产品长度进行计算的方法,然后以平均产品长度作为性能度量的标准讨论在不同生产情况下如何选择EWMA图的控制参数,从而获得最佳的监控效果。通过与通用的EWMA设计方法的比较,说明了本文方法更能快速地发现过程中出现的较小波动。
In this paper, we mainly discuss the optimal design of EWMA charts for processes monitoring. At first, a Markov chain method is proposed to calculate the average product length of EWMA chart. Then we use APL as a kind of criterion for measuring control chart performance to decide how to choose optimal parameters of this chart in different situations. By comparing with traditional EWMA design method, we find that this method can detect little shifts in processes more quickly.