Artificial Neural Network Prediction of Optimal Phenylic acid Adsorption using Lantana camara Activated Carbon

Abstract

This paper exemplifies the application of artificial neural network (ANN) for prediction of performances in adsorption of phenylic acid from waste water by conventional and low cost Lantana camara activated carbon as adsorbent material. To estimate the removal efficiencies of phenylic acid, a three-layer feed-forward neural network using a back propagation algorithm was utilised in the MATLAB environment. The initial concentrations (mg/L) of phenylic acid, amount (g/L) of adsorbent and pH are the input parameters utilised to train the neural network. The output of the neural network was taken to be the effectiveness of phenylic acid removal. Statistical measures like root mean square error and linear regression were also used to evaluate the effectiveness of the proposed ANN models. Based on the comparison of the removal efficiencies of contaminants using ANN models and empirical results, ANN modelling for the adsorption of phenolic compounds was found to be reasonably consistent with the empirical results.