Storing, retrieving, and analyzing experimental catalyticdata with the help of artificial intelligence methods

Within the framework of the development of a decision support system devoted to the prediction of the best catalyst and reaction conditions for industrially important aromatic reactions catalyzed by zeolites, we propose some techniques to derive correlations from experimental catalytic data stored in a knowledge base by means of artificial intelligence (AI) methods. Having collected miscellaneous experimental informations available from the literature regarding the alkylation of toluene and benzene with light alcohols and olefins, several tools have been developed to retrieve them easily and rapidly. Plotting graphs after several selections imposed by user's constraints provides conclusions on the catalytic behavior of the zeolite(s). This approach constitutes the first step on the development of an expert system for the prediction of a suitable catalytic system for a given type of reaction>[1]. © 1995 Elsevier B.V. All rights reserved.