Artificial Neural Network Modeling of Cr(VI) Biosorption from Aqueous Solutions

Abstract

Artificial neural network (ANN) model was applied for predicting the biosorption capacity of excess municipal wastewater sludge for hexavalent chromium (Cr(VI)) ions from aqueous solution. The effects of initial concentration (5 to 90 mg/L), adsorbent dosage (2 to 10 g/L), initial pH (2 to 8), agitation speed (50 to 200 rpm) and agitation time (5 to 480 min) were investigated. The maximum amount of chromium removal was about 96% in optimum conditions. The experimental results were simulated using ANN model. Levenberg-Marquardt algorithm was used for the training of this network with tangent sigmoid as transfer function at hidden and output layer with 13 and 1 neurons, respectively. The applied model successfully predicted Cr(VI) biosorption capacity. The average mean square error is 0.00401 and correlation coefficient between predicted removal rate and experimental results is 0.9833.