Prediction of MEUF process performance using artificial neural networks and ANFIS approaches

In the present study, a micellar-enhanced ultrafiltration (MEUF) procedure for the separation of lead ions from aqueous solution using response surface methodology (RSM) has been proposed. Due to the extreme complexity and nonlinearity of membrane separation processes, two models, including a feed forward artificial neural network (ANN) and an adaptive neuro-fuzzy inference system (ANFIS) have been utilized. These simulation methods have been given extreme accurate model that are more efficient than the second quadratic mathematical model for both response variables. The results of ANN and ANFIS models have been shown that the independent predicted rejection and permeate values were compared to measured target values and good correlations were found (R2 > 0.92, R2 > 0.97) for two above mentioned approaches, respectively.

► Pb (II) is a common heavy metal that even at low concentrations can be toxic.
► MEUF, a novel hybrid membrane separation process, is promising in removing metal.
► We have used BBD that do not require an excessive number of experimental runs.
► The BBD is either rotatable or nearly rotatable.
► The trained network using LM algorithm provides minimum error.
► In ANFIS, in comparison with using a single method enhanced prediction capabilities.