摘要
常规控制图应用的基本假设是从过程得到的观测值彼此独立。但许多过程出现了自相关现象。该文分别运用单值控制图和残差控制图就受控状况和失控状况的观测值对案例进行了分析比较。结果表明 :当过程存在自相关时 ,运用残差控制图更合适 ,但是 ,当自相关参数大于 0时 ,残差控制图检测过程异常的灵敏性有待提高。对于在现代生产过程中自相关数据 ,建议使用残差控制图 。
A basic assumption in standard applications of conventional control charts is that observations from the process at different times are independent random variables. But many manufacturing processes display inherently autocorrelated behavior. This paper introduced an individual control chart for the process mean and a residual chart to control autocorrelated observations from in-control and out-of-control processes. A case study shows that the residual chart is more suitable for autocorrelated process than the traditional control chart. The residual chart can detect out-of-control processes more quickly when the autocorrelated process parameter is larger than 0. For autocorrelated process, a residual chart is suggested instead of traditional control charts.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2002年第6期735-738,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金资助项目 ( 70 0 0 2 0 0 5 )
关键词
自相关过程
残差控制图
时间序列模型
autocorrelated process
control charts
time series analysis