Modelling, Optimization And Prediction Of Sorption Of Pb(II) And Mn(II) Ions From Wastewater Onto Acid Activated Shale Using Adaptive Neuro Fuzzy Inference Systems (ANFIS), Response Surface Methodology (RSM) and Modular Neural Network (MNN)

Authors : Idowu Ilaboya, Osadolor Izinyon

Pages : 1-22

View : 11 | Download : 12

Publication Date : 2020-03-27

Article Type : Research

Abstract :Batch experimental technique was employed to evaluate the effects of adsorption variables such as initial metal ion concentration, adsorbent dose, pH, and contact time on the sorption efficiency of Pb(II) and Mn(II) ions onto acid activated shale. To select the input variables with the highest significant contributions towards the sorption of Pb(II) and Mn(II) ions onto acid activated shale, adaptive neuro-fuzzy (ANFIS) was employed. Thereafter, statistical design of experiment (DOE) using central composite design was used to generate the data for modelling and prediction using a modular neural network (MNN). To produce accurate network architecture for prediction, the input data were first normalized to avoid the problems of weight variation. Thereafter, different training algorithm and hidden neurons were selected and tested to ascertain the optimum number of hidden neuron and the best training algorithm that will produce the most accurate network.   The linear coefficient of determination in addition to the mean square error for training and cross-validation was employed as the selection criteria. Results obtained shows that, Levenberg Marquardt Back Propagation training algorithm with 2 hidden neurons in the input and output layer with tangent sigmoid transfer function produced the most accurate prediction network. In addition, the modular neural network gave a strong agreement between the experimental and predicted sorption efficiency of Pb(II) and Mn(II) ions with R 2 values of 0.977 and 0.9648 having performance statistics of RMSE (0.03815), NRMSE (0.04097), Max.AE (0.02621), Min.AE (0.00041) and R 2 (0.988).
Keywords : Modular neural network, sensitivity analysis, Response surface methodology, central composite design