ECAI 2004 Conference Paper

[full paper]

Soft Computing Techniques applied to industrial catalysis

Soledad Valero, Estefania Argente, Jose Manuel Serra, Vicente Botti, Avelino Corma

A soft computing technique based on the combination of neural networks and a genetic algorithm has been developed for the discovery and optimisation of new catalytic materials when exploring a high-dimensional space. The genetic algorithm, based on real codification, allows to deal with different problems in chemical engineering. Also, the neural network simulates the reaction process and allows to calculate the fitness functions used by the genetic algorithm. One possible application of this technique is the optimisation of the catalytic performance of new solid materials by exploring simultaneously a big number of variables as elemental composition, manufacture procedure variables, etc. Another application is the optimisation of process conditions in catalytic reactors at industrial scale. Considering the high temporal and financial costs required for synthesizing and empirically testing potential solid catalysts, the application of soft computing techniques in this field seems really interesting, as the number of experiments could be reduced. The proposed system has been validated using two hypothetical functions, based on the modelled behaviour of multi-component catalyst explored in the field of combinatorial catalysis. Moreover, this soft computing technique has been applied to an industrial problem, being possible to obtain an optimised Ti-silicate catalyst for the epoxidation of olefins.

Keywords: soft computing, genetic algorithm, optimization, neural networks, combinatorial catalysis

Citation: Soledad Valero, Estefania Argente, Jose Manuel Serra, Vicente Botti, Avelino Corma: Soft Computing Techniques applied to industrial catalysis. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.765-769.