Catalytic reduction of NO by CO over CeO2-MOx (0.25) (M = Mn, Fe and Cu) mixed oxides—Modeling and optimization of catalyst preparation by hybrid ANN-GA

Abstract Catalytic performance of CeO 2 -MO x (0.25) (M = Mn, Fe and Cu) mixed oxide nanocatalysts were investigated in NO + CO reduction. Sol-gel method was used to synthesize nanocrystalline mixed oxides. Catalysts were characterized by XRD, BET, SEM, TEM and H 2 -TPR analysis. The Ce-Cu mixed oxide catalyst showed superior activity than other catalysts (with 80% NO and 72% CO conversions), due to its better reduction properties. To model and optimize the NO and CO conversions, a neuro-genetic approach was employed. This approach established by combining an artificial neural network with a genetic algorithm. The results showed that the ANN model is accurate with R 2  = 0.991, 0.979 and 0.960 for training, validation and testing, respectively. Catalyst design factors (Cu/Ce molar ratio, citric acid/nitrate and calcination temperature) were optimized by GA. The optimum values were 0.49, 0.98 and 500 °C. For Cu/Ce molar ratio, citric acid/nitrate and calcination temperature, correspondingly. NO conversion predicted through ANN-GA system and obtained via experimental at 300 °C were 91% and 90%, respectively.