▎ 摘　　要

The mesoporous Pd-Fe magnetic nanoparticles immobilized on the reduced graphene oxide were employed in the present work for the decolorization of toxic brilliant green in aqueous phase. The decolorization process was modeled using backpropagation artificial neural network and optimized by genetic algorithm and particle swarm optimization. These magnetic nanocomposites were synthesized by the two-step reaction in aqueous phase method and then characterized with various methods. According to response surface methodology, the effect of operating parameters on the decolorization of brilliant green in aqueous solution was studied through batch experiments. On the basis of these experiments, the prediction ability of response surface methodology and backpropagation neural network method was assessed. The decolorization process follows Freundlich isotherm and pseudo-second-order kinetics. Furthermore, thermodynamics studies demonstrate that the adsorption of brilliant green onto the nanocomposites was endothermic and spontaneous. Overall, these mesoporous nanomaterials have the advantages of strong adsorption capacity and fast decolorization for brilliant green, and modeling of the removal process with artificial neural network was successful.