A Neuro-Fuzzy Model for prediction of CaCl2 rejection in positively charged nanofiltration membrane

چکیده مقاله

A neuro-fuzzy modeling tool (ANFIS) has been used to model a composite NF membrane which had been fabricated by the modification of a polysulfone ultrafiltration support. The active top layer had been formed by the interfacial crosslinking polymerization of poly(ethylene imine) (PEI) with p-xylene dichloride (XDC). Then, it had been quaternized by methyl iodide (MI) to form a perpetually positively charged layer. This study aims to predict CaCl2 rejection in NF process as a function of the PEI solution concentration, PEI coating time, XDC concentration, crosslinking time, MI concentration and pH. Such processes show complex non-linear behavior. In this paper, ANFIS were applied to predict results. The results reveal that there is an excellent agreement between the tested and modeled data, with a good degree of accuracy.