Comparative Study Between Response Surface Methodology and Artificial Neural Network for Adsorption of Crystal Violet on Magnetic Activated Carbon

Abstract

The easily separable and regenerable magnetic activated carbon was synthesized for adsorption of toxic cationic dye, crystal violet, from aqueous solution. The synthesized magnetic activated carbon was characterized by SEM–EDX. The magnetic property of sorbent was evaluated by VSM method. The obtained saturation magnetization of 41.56emu g−1 showed facile separation of sorbent after adsorption process. The effect of five parameters of pH, temperature, time, initial dye concentration and sorbent amount on adsorption (%) were investigated. The percentage of adsorption was mathematically described as a function of experimental parameters and was estimated by central composite design (CCD). The maximum adsorption percent of 99.5±0.2 was obtained experimentally which was close to the percent of CCD prediction of 99.90 %. The same design was used for a three-layer artificial neural network model (ANN). The predicted data of CCD versus ANN showed the linear agreement with regression value \({(R^{2})}\) of 0.9994 which confirmed the ideality of CCD and ANN. The results of two models were compared in terms of coefficient of determination \({(R^{2})}\) and mean absolute percentage error (MAPE) to indicate the prediction potential of CCD and ANN. The MAPE (%) of 0.59 and 0.38 was found for CCD and ANN respectively. The obtained results indicated higher capability and accuracy of ANN in prediction. The experimental data were found to be properly fitted to the Langmuir and Freundlich models which indicates that the sorption takes place on a heterogeneous material and the sorption capacity of 12.59 mgg−1 was achieved.