Comparative Analysis of the Efficiency of Multi-Dimensional and Unbound One-Dimensional Control Charts Used for Process Indicator Change Tracking

摘要

The paper examines the design techniques of control charts intended for predicting critical time changes of lab-measured process indicators. A control chart design algorithm is offered for monitoring time changes of several product's properties with autocorrelations between consecutive observations. The algorithm is based on the filtration of observations by a specially designed filter without edge effect and on the Hotelling statistics.