Application of central composite design and artificial neural network in modeling of reactive blue 21 dye removal by photo-ozonation process

Abstract

The present study deals with use of central composite design (CCD) and artificial neural network (ANN) in modeling and optimization of reactive blue 21 (RB21) removal from aqueous media under photo-ozonation process. Four effective operational parameters (including: initial concentration of RB21, O3 concentration, UV light intensity and reaction time) were chosen and the experiments were designed by CCD based on response surface methodology (RSM). The obtained results from the CCD model were used in modeling the process by ANN. Under optimum condition (O3 concentration of 3.95 mg L-1, UV intensity of 20.5 W mr2, reaction time of 7.77 min and initial dye concentration of 40.21 mg L-1), RB21 removal efficiency reached to up 98.88%. A topology of ANN with a three-layer consisting of four input neurons, 14 hidden neurons and one output neuron was designed. The relative significance of each major factor was calculated based on the connection weights of the ANN model. Dye and ozone concentrations were the most important variables in the photoozonation of RB21, followed by reaction time and UV light intensity. The comparison of predicted values by CCD and ANN with experimental results showed that both methods were highly efficient in the modeling of the process.