Use of a neural-network-based approach for a reliable modelling of a distillation column
by Yahya Chetouani
International Journal of Modelling, Identification and Control (IJMIC), Vol. 11, No. 1/2, 2010

Abstract: Process development and continuous request for productivity led to an increasing complexity of industrial units. The main aim of this paper is to establish a reliable model of process behaviour both for the steady-state and unsteady-state regimes. Multilayer feedforward neural network in the form of non-linear-auto-regressive with exogenous input (MFNN-NARX) was proposed to predict the product quality of a distillation column. This study shows also another technique for neural model reduction into account the physical knowledge of the process. It shows the choice and the performance of the neural network in the training and test phases. An analysis of the inputs choice, time delay, hidden neurons and their influence on the behaviour of the neural estimator is carried out. Three statistical criteria are used for the validation of the experimental data. Based on these results, the best network MFNN-NARX is composed of one hidden layer of eight nodes and six exogenous inputs. This result shows the good generalisation of the developed reliable model.

Online publication date: Mon, 20-Sep-2010

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Modelling, Identification and Control (IJMIC):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com