PCA based statistical process monitoring of grinding process

Multivariate statistical process monitoring (MSPM) has received increasing attention, which is applied to improve process operations by detecting when abnormal process operations exist and diagnosing the sources of the abnormalities. This paper presents a MSPM application method on grinding processes, including principal component analysis (PCA), fault detection and fault diagnosis using the contributions from squared prediction error (SPE) statistic, and utilizes actual process data for verifying the validity of the method.