Modeling and optimization of toluene oxidation over perovskite-type nanocatalysts using a hybrid artificial neural network-genetic algorithm method

Toluene oxidation activities of sol-gel synthesized La1-xCexMn1-yCuyO3 perovskite-type catalysts were modeled and optimized using an intelligent approach. To design an intelligent system, an artificial neural network was coupled with a genetic algorithm. Catalyst formulation (mole fractions of Ce and Cu) and calcination temperature were optimized to have enhanced toluene conversion. The results showed that the best neural network architecture could predict toluene oxidation at an acceptable level of accuracy. The model prediction results indicated the maximum toluene conversion to be produced by La1-xCexMn1-yCuyO3 of the following formulation details: Ce mole fraction of 0.30, Cu mole fraction of 0.52, and calcination temperature of 625 degrees C. The optimized values of toluene conversion obtained via prediction model and experimentations at 240 degrees C were 92.7% and 92.1%, respectively. The prepared perovslcite catalysts were characterized by XRD, BET, H-2-TPR, and SEM. (C) 2016 Published by Elsevier B.V. on behalf of Taiwan Institute of Chemical Engineers.