具有可变抽样区间的非正态累积和控制图

CUSUM Control Chart with Variable Sampling Intervals Under non-normal Distribution

当过程存在小波动时,累积和控制图比传统的休哈特控制图监控效果灵敏。为了提高控制图的监控效率,本文针对非正态情形下的累积和控制图进行可变抽样区间设计。首先用Burr分布近似各种非正态分布,构造可变抽样区间的非正态累积和控制图;其次利用马尓可夫链方法计算其平均报警时间;最后研究结果表明,所设计的可变抽样区间非正态累积和控制图较固定抽样区间的非正态累积和控制图能更好地监控过程的变化。

When the process has the small shift, CUSUM control chart can find the shift faster than the traditional Shewhart control chart. In order to improve the monitoring efficiency of control chart, CUSUM control chart with variable sample intervals（VSI）for non-normal data is proposed in this paper. Firstly, VSI CUSUM control chart is developed using Burr distribution to approximate various non-normal distributions. Secondly, the Markov chain method is used to calculate the average time to signal. Finally, the computing resuhs show that the VSI CUSUM control chart for non-normal data is more efficient in detecting shifts than the fixed sampling interval （FSI） CUSUM control chart for non-normal data.