Forecastable Component Analysis and Partial Least Squares Applied on Process Monitoring

DOI

10.2991/isci-15.2015.111How to use a DOI?

Keywords

Forecastable Component Analysis; Partial Least Squares; Fault Detection; TE Process;

Abstract

Forecastable Component Analysis (ForeCA) is a new feature extraction method for multivariate time series. ForeCA can find an optimal transformation to dig out the potential forecastable information structure from large amounts of data. This paper combines ForeCA with PLS for industrial process monitoring. This method overcomes the drawback that partial least squares(PLS) rarely use dynamic timing characteristics of system, so it can reflect the dynamic nature of industrial processes better. We use PLS for regression after appropriate forecastable components selected. Finally, we construct CUSUM statistic and SPE statistic for monitoring industrial processes. Simulation results on the Tennessee Eastman (TE) process illustrate the effectiveness of the proposed method for detecting slow drift fault.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Dan Wang
AU  - Yirong Lu
AU  - Yupu Yang
PY  - 2015/01
DA  - 2015/01
TI  - Forecastable Component Analysis and Partial Least Squares Applied on Process Monitoring
BT  - Proceedings of the 2015 International Symposium on Computers & Informatics
PB  - Atlantis Press
SP  - 839
EP  - 846
SN  - 2352-538X
UR  - https://doi.org/10.2991/isci-15.2015.111
DO  - 10.2991/isci-15.2015.111
ID  - Wang2015/01
ER  -