Applicability of ANN in Adsorptive Removal of Cd(II) from Aqueous Solution

Abstract

Cadmium is frequently used and is extremely toxic in relatively low dosages and is one of the principal heavy metals that is responsible for causing kidney damage, high blood pressure, renal disorder, destruction of red blood Cells and bone fracture. Permissible limit to discharge in the inland surface water is 2.0 mg/l, discharge in public sewers is 1.0 mg/l and drinking water is 0.01 mg/l. Adsorption is the only user-friendly technique for the removal of heavy metal. We have developed an ANN model for prediction of percentage removal of Cd(II). A multilayer perceptron with a single hidden layer has been learnt separately by three different algorithms: Backpropagation, Levenberg-Marquardt and Scaled Conjugate Gradient algorithms for analysis purpose. Optimization for each one of the four standard transfer functions (in a single hidden layer) has been carried out in all three cases. The ANN model with Backpropagation algorithm, with the second transfer function and 25 processing elements gives the best predictability of the outlet concentration.