Robust PCA and Normal Region in Multivariate Statistical Process Monitoring

摘要

The on-line monitoring and evaluation of process operating performance are essential to safe operation and consistent high quality production. This can be achieved using multivariate statistical approaches (MSA). The basic philosophy of MSA is that the behavior of the process is characterized by data obtained when the process is operating well or in normal region. Subsequently, future unusual events can be detected by referring the measured process behavior against this normal region model, i.e., normal region is used here as a calibration criteria of the process.