Catalyst Development Using Soft Computing and High Throughput Screening (4) Activity Mapping of Copper Catalyst for Methanol Synthesis by Neural Network

摘要

Simple Genetic algorithm (GA) was applied for simulation in optimization of the Cu/Zn/Al/Sc ratio of mixed oxide catalyst for methanol synthesis from syngas. The simulated activity, calculated by equations derived from experimental results, was used instead of experimental results through out this study to evaluate the fittness in GA program. The neural network (NN), trained by the 92 data determined by GA program, successfully mapped the catalytic activity. The trained NN then was used to evaluate the fitness of the catalyst code instead of experimental steps. This substitution can eliminate laborious steps, even though parallelized, such as catalyst preparation and activity test, from the optimization loop. The combination of catalyst design by GA and the activity evaluation by NN is expected to be promising for high efficient catalyst screening.