Design of a propane ammoxidation catalyst using artificial neural networksand genetic algorithms

Artificial neural networks (ANNs) and genetic algorithms (GAs) are applied to the optima design of a catalyst for propane ammoxidation, The mole perce ntages of six components of a catalyst (P, K, Cr, Mo, Al2O3/SiO2, and VSb5W Sn) are used as inputs, and the activity and the acrylonitrile selectivity serve as the two outputs. This trained optimal linear combination (OLC) net work is used to evaluate the yield of new catalyst compositions generated d uring GA optimization. The best yield of acrylonitrile found after GA optim ization is 79%, which is higher than the highest yield previously reported (64%). The OLC neural network, using the acrylonitrile yield (i.e., activit y times selectivity) as the output, greatly improves the simulation of the catalyst system compared to a simple, single-network architecture. In parti cular, whereas single-network methods can all easily reproduce the experime ntal patterns used for training and validation, the OLC is markedly superio r for generalizing to novel catalyst patterns.