Abstract
An immune algorithm (IA), which is a heuristic method for selection, was used to optimize compositions of the Cu/Al2O3 catalyst for the methanol steam-reforming reaction. An artificial neural network (ANN) that was trained on observed data was adopted to evaluate the catalytic activity in an immune algorithm. After 10,000 generations, we obtained several candidate catalysts. Compared with the results of a genetic algorithm (GA) for the same evaluation function, the results of an IA presented potentially useful candidate catalysts. During initial stages of developing catalysts, extensive screening using search methods, such as immune algorithms, should prove effective.
