Modeling of effluent COD in UAF reactor treating cyanide containing wastewater using artificial neural network approaches

Abstract

In this study the performance of the upflow anaerobic filter (UAF) reactor treating cyanide was simulated using three different neural network techniques (ANNs) - multi-layer perceptron (MLP) neural network, radial basis neural network (RBNN), and generalized regression neural network (GRNN). The performance of UAF reactor over a period of 130days at different cyanide concentrations was evaluated with these robust models. Influent chemical oxygen demand (CODin), hydraulic retention time (HRT), and influent cyanide concentration (CNin) were the inputs of the models, whereas the output variable was effluent chemical oxygen demand (CODeff). The models' results were compared with each other using four statistical criteria - root mean square error (RMSE), mean absolute error (MAE), mean absolute relative error (MARE), and determination coefficient (R^2). The results showed that the MLP neural network with Levenberg-Marquardt algorithm was found to be better than the RBNN and GRNN techniques.