多元自相关过程的VAR控制图

为了解决多元自相关过程的残差T~2控制图对小偏移不灵敏的问题,本文利用批量-均值法的思想,结合VAR模型的渐近分布,设计了多元自相关过程的向量自回归(VAR)控制图.只要子组样本量足够大,VAR控制图可以对过程出现的各种偏移进行有效控制.通过对比残差T~2控制图的控制效果,得出VAR控制图对小偏移灵敏、残差T~2控制图对大偏移灵敏的结论,联合使用VAR控制图和残差T~2控制图可更有效地监控多元自相关过程。

A VAR Control Scheme for Multivariate Autocorrelated Observations

Residual-based T~2 control chart is not effective for monitoring the multivariate autocorrelated processes with shifts extremely small.In this paper,based on the thinking of batch-means,a control procedure is provided using the methodological conclusion of distribution characteristics for vector au- toregressive model(VAR),termed the VAR control scheme.As long as subgroup size is sufficient large, VAR control scheme can be used efficiently to monitor all shifts,even extremely small.Compared with residual-based T~2 control chart,VAR control scheme is sensitive for small shifts and residual-based T~2 control chart is sensitive for big shifts comparatively.Thus it is more effective to monitoring the multi- variate autocorrelated observations by combining VAR control scheme and residual T~2 control chart.