Synthesis of T2 and Q statistics for process monitoring

Abstract

In this work, a novel approach that synthesizes the T² and Q statistics for statistical process condition monitoring is introduced. The synthesis is based on the joint probability density function of the two statistics, which is estimated with kernel density estimation. This inherits the following benefits: (i) it can be more sensitive to detect abnormal process behaviour than the individual statistics and (ii) it reduces the number of monitoring charts to be observed. The utility of this new approach is demonstrated using two application studies that involve the simulation of a fluid catalytic cracking unit and recorded data from an industrial distillation unit.