Development of an artificial neural network model for adsorption and photocatalysis of reactive dye on TiO2 surface

Development of an automated wastewater treatment plant is very difficult as the parameters of an industrial effluent change severely; accordingly the change in output of treatment plant. A computer-simulated model is required for interrelating the input and output parameters of wastewater treatment plant. An artificial neural network model has been proposed for the prediction of adsorption and photocatalysis efficiency of TiO2 photocatalyst. The network was trained using the experimental data obtained at different pH with different TiO2 dose and initial dye concentration. Different algorithms and transfer functions for hidden layer have been tested to find the most suitable and reliable network. The optimum number of neurons in the hidden layer was found by trial and error method. These neural network models efficiently predict the adsorption efficiency (% dye removal), adsorption capacity (loading) and photocatalytic efficiency of the process. Solution of reactive black 5 was used as simulated dye wastewater for this study. The effect of different operating parameters on process efficiency was studied.