Comparative Study of Artificial Neural Networks (ANN) and Statistical Methods for Predicting the Performance of Ultrafiltration Process in the Milk Industry

• Main Authors: Javad Sargolzaei, Naser Saghatoleslami, Sayed Mohammad Mosavi, Mohammad Khoshnoodi
• Format: Article
• Language: English
• Published: Iranian Institute of Research and Development in Chemical Industries (IRDCI)-ACECR 2006-06-01
• Series: Iranian Journal of Chemistry & Chemical Engineering
• Subjects: milk ultrafiltration; artificial neural networks; statistical methods; permeate flux; hydraulic resistances
• Online Access: http://www.ijcce.ac.ir/article_8092_3e23d617a8613ad2cfc7553cd4fc9003.pdf

Milk ultrafiltration is a membrane process, which is highly complex innature. The cost effectiveness of the process depends heavily on the flux permeate and the total hydraulic resistance of the membrane. In this work, a comparative study for the prediction of the performance of milk ultrafiltration with ANN and statistical method has been carried out. The result reveals that both methods carry out the prediction with a high degree of accuracy. However, the statistical method, contrary to neural nets, is both costly and time consuming and the accuracy of the data are also in doubt, as the operating conditions are not consistent throughout each of the test runs. The result also reveals that there is a good agreement between the predicted fluxes permeates and the total resistances of this work with the actual values. The findings of this study also shows that the artificial neural nets technique can be applied as a powerful tool and a cost and time effective way in predicting and assessing the performance of  milk ultrafiltration process.