Variable Selection based Fault Isolation Strategy for Multivariate Statistical Process Monitoring

Authors

  • Sumit Mohanty , Satheesh Kumar J

Abstract

Multivariate Statistical Process Monitoring(MSPM)is used normally in industrial processes
where large number of process variables are involved and complex correlations exists between the process
variables. In this paper an efficient fault isolation based process monitoring method is discussed using
variable selection. Sparse Partial Least Square(SPLS) is a special type of discriminant analysis for doing
variable selection based on the classification between normal process data and actual process measurement
data. The SPLS based variable selection for fault isolation is explained here by implementing with real time
industrial process data.

Published

2020-04-30

Issue

Section

Articles