Modeling of catalyst composition–activity relationship of supported catalysts in NH3–NO-SCR process using artificial neural network

This paper presents an artificial neural network (ANN) for modeling the relationship between catalyst composition and catalytic performance in the NH3-SCR of NO process. The supported catalysts with different transition metals (Mn, Fe, Co and Cu) and (γ-Al2O3, ZSM5 and SAPO-34) supports were prepared and tested in NH3–NO-SCR reaction to generate required data for neural network development. The ANN was constructed using the experimental dataset, and all the data were integrated using support and metal atomic descriptors for the construction of general catalyst design model. The statistical analysis of the results indicated that the R2 values for the training and test data were high, more than 0.9, and this indicates that ANN-based model developed in this work can predict catalyst performance correctly. More evaluation of the obtained model revealed that metal has more influence than support on catalyst activity at supported catalysts.