Multivariate Statistical Approaches to the Analysis, Monitoring
and Diagnosis of Process Operations
Tuesday 3 June, 9:00--10:00
- Organizer: Scott VANDER WIEL
- Bell Labs
- Moderator: William MEEKER
- Iowa State University
- John F. MACGREGOR
- McMaster University, Hamilton, Ontario, Canada
Process computers routinely collect large amounts of data from a
multitude of on-line sensors. Most of these data are never used in
any effective way. This presentation will give an overview of some
multivariate statistical approaches for the analysis of these data and
for the on-line monitoring and diagnosis of process operations. These
approaches are based on multivariate statistical projection methods
such as Principal Component Analysis (PCA) and Projection to Latent
Structures (PLS) which are capable of compressing the data down into
low dimensional spaces where the analysis and interpretation are much
easier.
Justification for using these multivariate approaches to SPC will be
given, and the methods will be illustrated with industrial
applications to both batch and continuous processes.
[John F. MacGregor, McMaster University, Hamilton, Ontario, L8S 4L7, Canada, macgreg@mcmaster.ca]
Back to JRC program
Back to JRC Home Page