Prediction of azo dye decolorization by UV/H2O2 using artificial neural networks

An artificial neural network model was developed to predict the photochemical decolorization of C.I. Acid Orange 7 solution by a combination of UV and hydrogen peroxide. The initial concentrations of dye and hydrogen peroxide, the pH of the solution and time of UV irradiation were employed as input to the network; the output of the network was decolorization efficiency. The data used in this study were obtained from our previous papers. The multilayer feed-forward networks were trained by 114 sets of input–output patterns using a backpropagation algorithm; a three-layered network with eight neurons in the hidden layer gave optimal results. The model gave good predictions of high correlation coefficient (R2 = 0.996). As expected, the initial concentration of H2O2 with a relative importance of 48.89%, appeared to be the most influential parameter in the decolorization process.