MSPC-Multivariate Statistical Process Control

Course Description

Today’s highly instrumented chemical and manufacturing processes produce a tremendous amount of data, much of which is archived and only reviewed after a major process upset or fault. MSPC-Multivariate Statistical Process Control covers methods and strategies for dealing with this data overload and extracting critical information about process health. The course covers monitoring and fault detection in chemical and manufacturing processes. Methods for monitoring continuous, batch and transient processes are covered. Using diagnostic plot to track down root causes is covered, along with methods for dealing with process drift. The course includes hands-on computer time for participants to work example problems using PLS_Toolbox.

Prerequisites

Linear Algebra for Chemometricians, MATLAB for Chemometricians or equivalent experience. Chemometrics I–PCA or equivalent experience highly recommended.

Course Outline

1. General principles of SPC and fault detection
2. A favorite tool: Principal Components Analysis
   Some examples of PCA for MSPC
3. Diagnostic Plots for Interpreting and Sourcing Faults
   PCA Scores and Loadings
   Q and T2 Statistics
   Contribution plots
4. Theoretical basis for MSPC 
   Time Series Models and Lagged Variables
   More examples
5. Monitoring Batch Processes-Multi-way Models
   Unfold PCA (aka Multi-way PCA)
   PARAFAC and Tucker Models
   Comparison of methods on some example data
6. Dealing with process drift
   Examples
7. Conclusions
   
8. Additional examples and homework