Dey, Rahul
(2016)
Process fault detection and diagnosis of fed-batch plant using multiway principal component analysis.
Masters thesis, Indian Institute of Technology Hyderabad.
Abstract
With the advent of new technologies, process plants whether it be continuous or batch process
plants are getting complex. And modelling them mathematical is a herculean task. Model based
fault detection and diagnosis mainly depend on explicit mathematical model of process plant, which
is the biggest problem with the model based approach. Whereas with process history based there is
no need of explicit model of the plant. It only depends on the data of previous runs.
With the advancement in electronic instrumentation, we can get large amount of data electronically.
But the crude data we get is not useful for taking any decision. So we need develop techniques
which can convey us the information about the ongoing process. So we take the help of multivariate
statistics such as Principal Component Analysis(PCA) or Partial Least Squares(PLS). These
methods exploits the facts such as the process data are highly correlated and have large dimensions,
due to which we can compress them to lower dimension space. By examining the data in the lower
dimensional space we can monitor the plant and can detect fault.
Actions (login required)
 |
View Item |
Altmetric
Altmetric