BP-ANN Approach for Modeling Cd(II) Bio-Sorption from Aqueous Solutions Using Cajanus cajan Husk

Abstract

This work aims at the modeling of bio-sorption of cadmium(II) onto physically and chemically activated Cajanus cajan (Pigeon pea) husks. Experimental data obtained were fitted to a number of isotherm and kinetic models, and the results interpreted. The monolayer Cd(II) bio-sorption capacities of the husk were found to considerably increase by 2.82 times due to chemical activation, for bio-sorption from a solution containing an initial Cd(II) concentration of 100 mg/L and by about 1.78 times for a solution containing an initial Cd(II) concentration of 150 mg/L. Further, the BackPropagation Artificial Neural Network (BP-ANN) was applied to understand the accuracy and prediction of isotherm and kinetic data. The tangent sigmoid transfer function was used at the input to hidden layer whereas a linear function was used at output layer. The isotherm and kinetic data were distributed into training (65%) and testing (35%) phase. The training, testing, and prediction by BP-ANN were found to be adequate, with an absolute relative percentage error of 2.1827 and correlation coefficient R2of 0.9967 and 0.9863 at prediction for isotherm and kinetic studies, respectively. Comparison of BP-ANN and experimental results indicated that the prediction model is capable of predicting the bio-sorption effectiveness with good accuracy.