自相关过程的采样间隔与控制效果

化工行业中,一些过程数据存在明显的自相关,不满足统计过程控制理论中观测值相互独立的假定。该文以化工行业中的常见设备——储液罐为例,对采样间隔与控制图的控制效果之间的关系进行探讨。以平均运行时间为新的评价标准,得出结论:在满足稳态平均运行时间不变的情况下,采样间隔越短,控制图越能较快地检测出过程均值的异常,即缩短采样间隔可以改进控制效果。

Sampling interval and performance of control charts for autocorrelated processes

In the chemical industry, observations from many processes are autocorrelated so they fail to satisfy the fundamental assumption of independence for traditional statistical process control. The analysis of the autocorrelated events was based on a simple stirred tank which is sufficiently simple yet physically representative. The relationship between the sampling interval and the performance of control charts was analyzed using the average run time as a comparison standard. The fixed in-control average run time should be kept constant since the out-of-control average run time will decrease if the sampling interval is shortened, so that the out-of- control state can be detected more quickly. The sampling interval can be shortened to improve the control chart performance.