Improved multi-scale principal components analysis with applications to process monitoring

Abstract

Multi-scale monitoring approaches, which combine principal component analysis (PCA) and wavelet analysis, have received considerable attention. These approaches are potentially very efficient for detecting and analyzing diverse ranges of faults and disturbances in chemical processes. An improved multi-scale principal component analysis (MSPCA) is proposed for polymerization process monitoring. Improved MSPCA simultaneously extracts both, cross correlation across the variable and auto-correlation within a variable. Using wavelets, the individual variable are decomposed into approximations and details at different scales. Contributions from each scale are collected in separate matrices, and a PCA model is then constructed to extract correlation at each scale. The proposed improved multi-scale method is successfully applied into fault detection of polymerization process.