Multivariate Statistical Process Control (MSPC) and Batch SPC (BSPC)
Course Description
Today’s highly instrumented manufacturing processes produce tremendous amount of data, much of which is archived and reviewed only after a major process upset or fault. Multivariate Statistical Process Control (MSPC) and Batch Statistical Process Control (BSPC) provide useful tools for proactive process monitoring, fault detection and fault diagnosis resulting in higher productivity and efficiency applicable to chemical, pharmaceutical, semiconductor and other manufacturing endeavors. The course introduces MSPC methods and strategies for dealing with data overload in continuous processes and extracting critical information about process health. The concepts are then extended to batch processes. The course includes hands-on computer time for participants to work example problems using PLS_Toolbox and Solo software.
Prerequisites
Linear Algebra for Chemometricians, MATLAB for Chemometricians or equivalent experience. Chemometrics I–PCA or equivalent experience highly recommended.
Course Outline
- Introduction
General Principles of SPC and Fault Detection
Use of Models
The Multivariate Advantage - A favorite tool: Principal Components Analysis
Some examples of PCA for MSPC
Diagnostic Plots for Interpreting and Sourcing Faults
PCA Scores and Loadings
Q and T2 Statistics
Contribution plots - Theoretical basis for MSPC
Time Series Models and Lagged Variables
More examples - Modeling Batch Processes
Unfold PCA (aka Multi-way PCA)
Summary PCA (SPCA)
Batch Maturity Index Models
Multi-way Models: Parallel Factor Analysis (PARAFAC) and PARAFAC2
Comparison of models on example data - Strategies for Dealing with Unequal Batch Record Length
Summary Variables
Correlation Optimized Warping (COW)
Dynamic Time Warping (DTW)
Extent of Reaction and other Indicator Variables
Batch Maturity Index
Comparison of methods on some example data - Dealing with Process Drift
- Conclusions