ePrints

Enhanced modeling of an industrial fermentation process through data fusion techniques

Lookup NU author(s): Dr Sophia Triadaphillou, Professor Elaine Martin, Professor Gary Montague

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.

Abstract

A novel strategy for the analysis and interpretation of spectral data from a fermentation process is considered. The interpretation is challenging as a consequence of the large number of correlated spectral measurements recorded from the process in which a complex series of biochemical reactions occur. A full spectral analysis using PLS is the standard interpretation strategy. However, within this paper an alternative method, Spectral Window Selection (SWS), is proposed, and compared with that of genetic algorithms. SWS is shown to provide a more robust calibration model. Furthermore its performance is hypothesised to be enhanced by multiple model bagging. This claim is investigated and proven. Finally an overall calibration model is compared with a local modelling approach. The methodologies are applied and compared on an industrial NIR data-set from an antibiotic production process. © 2005 Elsevier B.V. All rights reserved.

Publication metadata

Author(s): Triadaphillou S, Martin E, Montague G, Jeffkins P, Stimpson S, Nordon A

Publication type: Article

Publication status: Published

Journal: Computer Aided Chemical Engineering

Year: 2005

Volume: 20

Issue: C

Pages: 1393-1398

ISSN (print): 1570-7946

ISSN (electronic):

Publisher: Elsevier BV

URL: http://dx.doi.org/10.1016/S1570-7946(05)80074-4

DOI: 10.1016/S1570-7946(05)80074-4

Altmetrics

Share