Optimization of catalyst for methanol synthesis by a combinatorial approach using a parallel activity test and genetic algorithm assisted by a neural network

Optimization of a Cu-Zn-Al-Sc oxide catalyst for methanol synthesis from syngas was performed by a combinatorial approach using a genetic algorithm (GA) with/without various neural networks (NNs). Performance of optimum catalysts found by the various methods was compared. The catalyst with maximum activity was found by the combination of GA and radial basis function network (RBFN) in the optimization of Cu-Zn-Al-Sc oxide catalyst composition. Therefore this method was found to be the most efficacious method. In addition, we conducted the optimization on the RBFN with a larger population in the GA program to find the best catalyst in the early stage of evolution. On the other hand, we also tried to optimize simultaneously both composition and calcination temperature of a Cu-Zn oxide catalyst. In that case, the optimum catalyst was found by the combination of GA and back-propagation network. Thus, GA is a more robust tool when it is combined with NNs.