Modeling of milk ultrafiltration permeate flux under various operating conditions and physicochemical properties using Nero- Fuzzy method

In this study, an adaptive neuro-fuzzy inference system (ANFIS) used for the prediction of permeate flux as a function of the physico-chemical and operating parameters during ultrafiltration of milk. An ultrafiltration pilot plant equipped with hollow fiber module and polyethersulfone membrane (MWCO 10 kDa) was used to do the milk ultrafiltration with various physico-chemical properties, consists of five levels of pH (5.6, 6, 6.6, 6.9 and 7.6) and three levels of ionic strength (0.03, 0.06 and 0.12) and under different operating conditions including transmembrane pressure (TMP) at three levels (0.1, 0.3 and 1 atm), temperature at three levels (30, 40 and 50 ° C) and the flow rate at three levels (10, 30 and 46 m/s). In order to model the effects of operating parameters and physicochemical properties of milk on permeate flux, the experimental data was randomized. 30 % of the data for learning, 30% of the data for evaluation and 40 % of the data was used to test the model. The results showed that the Nero–Fuzzy modeling approach is capable to predict the permeate flux under various operating conditions and physiochemical characteristics of milk, and modeling results represented there was an excellent correlation (average R = 0.93) between the predicted data and experimental data.