Neural computation to predict TiO2 photocatalytic efficiency for nitrogen oxides removal

An artificial neural network (ANN) was built to analyze the decrease of nitrogen oxides relative to TiO2 powder photocatalytic efficiency. Experimental sets were considered to train and test ANN where it was possible to identify an optimized structure representing the correlations between experimental variables and NO and NOx efficiencies. Predicted results permitted to point out the role of exposed surface and powder mass.