Resumen de Multivariate Process Variability Monitoring Through Projection

Shuguang Hao, Shiyu Zhou, Yu Ding

• A linear process model is often built to connect the multivariate quality measurements with a set of fixed assignable causes. The column space of this model provides a subspace for projection and subsequent monitoring of a shift in the mean. However, such a projection may not be suitable for monitoring variability. An alternative projection statistic, called VS, is proposed for constructing a multivariate variability monitoring chart. The VS chart has less restrictive requirements on the process model and improved detection when compared with competing methods.