Application of partial correlation in active statistical process control

Abstract

Most of the research in Stastical Process Control (SPC) has been focused on the charting techniques which are employed to monitor the process. Unfortunately, little attention is focused on the importance of bringing the process control automatically via these charting techniques. By utilising Automatic Process Control (APC) concept it is possible to devise a method whereby the process is monitored and automatically controlled via SPC procedures. The proposed method avoids the time series or dynamic model identification problems, by focusing only on statistical input-output relationship via Partial Correlation Analysis (PCorrA). The advantage of this method lies in the ability of PCorrA to quantify the relationship between quality variables and input variables. A discussion of how to implement PCorrA in SPC is given. Finally the results of applying this strategy to a simulated reversible reaction process are shown and discussed.