Title: Model-order reduction based on artificial neural networks for accurate prediction of the product quality in a distillation column

Authors: Yahya Chetouani

Addresses: Departement Genie Chimique, Universite de Rouen, Rue Lavoisier 76821, Mont Saint Aignan Cedex, France

Abstract: The main aim of this paper is to establish a reliable model of a process behaviour both for the steady-state and unsteady-state regimes. The use of this accurate model allows distinguishing a normal mode from an abnormal one. Therefore, the neural black-box identification by means of a non-linear auto-regressive with exogenous model has been chosen. This study shows another technique for neural model reduction into account the physical knowledge of the process. An analysis of the inputs choice, time delay, hidden neurons and their influence on the behaviour of the neural estimator is carried out. After describing the system architecture, a realistic and complex application as a distillation column is presented in order to illustrate the reliability of the prediction and model reduction. Satisfactory agreement between identified and experimental data is found and results show that the neural model successfully predicts the evolution of the product composition.

Keywords: distillation columns; modelling; NARX; nonlinear auto-regressive with exogenous; artifical neural networks; ANNs; product composition; model reliability; product quality; prediction.

DOI: 10.1504/IJAAC.2009.026780

International Journal of Automation and Control, 2009 Vol.3 No.4, pp.332 - 351

Received: 18 Sep 2008
Accepted: 22 Dec 2008
Published online: 26 Jun 2009 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article