FEED-FORWARD BACK-PROPAGATION NEURAL NETWORK FOR THE ELECTRO-OXIDATION OF DISTILLERY EFFLUENT

The aim of the present work is to demonstrate the technical feasibility of treating high-strength distillery wastewater in an electrochemical flow reactor and to predict the result using an artificial neural network (ANN) model. The experiments were conducted using oxide coated on expanded titanium (Ti/Ru_0.3Ti_0.7O₂) as anode and stainless steel as cathode in a batch reactor with electrolytic recirculation. Pollutant degradation was measured as chemical oxygen demand (COD) for various operating parameters such as effluent flow rate, current density, and supporting electrolyte concentration. Experiments were conducted for various flow rates, supporting electrolyte concentrations, and current density. An ANN was used for modeling the experimental results. The model was developed using a feed-forward back-propagation network with different layers and neurons. It can be concluded that the network configuration of 3-3-3-1 simulation gives the best result in predicting the experimental results with a high correlation coefficient (R ² = 0.9987). Using this network model, the effluent COD removal can be predicted quickly and easily.