Removal of methylene blue via bioinspired catecholamine/starch superadsorbent and the efficiency prediction by response surface methodology and artificial neural network-particle swarm optimization

Abstract

This paper demonstrates coupling of the artificial neural network (ANN) technique with the particle swarm optimization (PSO) method and compares the performance of ANN-PSO with response surface methodology (RSM) in prediction of the adsorption of methylene blue (MB) by a novel bio-superadsorbent. To this, a starch-based superadsorbent was synthesized using acrylic acid and acryl amid polymers and then catecholamine functional groups were combined onto the surface with oxidative polymerization of dopamine. The adsorption of MB was considered as a function of pH, dye concentration, and contact time. The best topology of the ANN was found to be 3-7-1, and prediction model of the adsorption capacity was demonstrated as a matrix of explicit equations. ANN-PSO is more accurate than RSM. The results revealed that the root-mean-square error, correlation coefficient, and normalized standard deviation for the ANN-PSO are 22.46, 0.99, and 16.83, respectively, while for RSM are 82.89, 0.98, and 65.41, respectively.