A rapid and efficient sono-chemistry process for removal of pollutant: Statistical modeling study. (1st October 2019)

• Record Type: Journal Article
• Authors: Mossavi, E.; Sabzevari, M. Hosseini; Ghaedi, M.; Azqhandi, M.H. Ahmadi; Hosseini, S.J.
• Is Part Of: Polyhedron. Volume 171(2019)
• Page Start: 65
• Page End: 76
• Publication Date: 2019-10-01
• Subjects: Nanocomposite -- MWCNT/TiO2@Ag/Fe -- Experimental design -- ANFIS -- GRNN; Chemistry, Inorganic -- Periodicals; Chimie inorganique -- Périodiques; Organometaalverbindingen; Anorganische chemie; 546.05
• DOI: 10.1016/j.poly.2019.07.002
• Languages: English
• ISSNs: 0277-5387

Abstract

Graphical abstract: Abstract: In this research paper, Ag and Fe doped into TiO2 loaded on the Multi wall carbon nanotube (MWCNT/TiO2 @Ag/Fe-NC) was prepared and characterized. The adsorption efficiency was modeled by ANFIS (Adaptive Neuro-Fuzzy Inference System), GRNN (Generalized Regression Neural Network) and RSM (response surface methodology). The effect of process factors i.e. sonication time, the concentration of Methylene Blue (MB), MWCNT/TiO2 @Ag/Fe-NC mass and pH on the decolorization of MB was investigated by the RSM, GRRN, and ANFIS. The ability of all three models was examined by four statistical visualization such as R 2, RMSE (root mean square error), MAE (mean absolute error) and %AAD (absolute average deviation). The statistical visualization result for the validation dataset shows that the proposed approaches (i.e. ANFIS, GRNN and RSM) will be able to predicate and model the removal MB. Nevertheless, from obtained result it clear that the ANFIS approach has more precise in respect to the other models. Though, it was known that the Generalized Regression Neural Network is easier and take a little time for modeling than the Adaptive Neuro-Fuzzy Inference System approach. Therefore, the GRNN algorithm can be built a new prospect in predication and/or modeling and is also feasible that could be applied in actual systems.