Expert systems approach to computer-aided design of catalysts

Abstract

A catalyst design expert system, INCAP (INtegration of Catalyst Activity Patterns), for the selection of catalyst components was developed by decomposing the catalyst design problem into sub-problems to be solved concurrently and by building up a novel knowledge representation scheme for the integration of activity patterns. INCAP gives candidates for catalyst components through the knowledge-based reasoning of (1) the reaction mechanism of the target reaction, (2) catalytic functions required for the target reaction from the reaction mechanism, (3) plausible side-reactions from the reactants and main products, (4) unfavourable catalytic functions from the plausible side-reactions and (5) catalyst components by integrating various kinds of activity patterns data for the catalytic functions estimated in (2) and (4). Although the knowledge for the estimations of (1)–(4) could be encoded into a computer by using the conventional knowledge representation of the production system and/or frame system, a new representation scheme had to be developed for that of (5). The whole system was written in Prolog language. INCAP was applied to the design of the catalyst for the oxidative dehydrogenation of ethylbenzene to examine the feasibility of the expert systems approach to design of catalysts. The results strongly suggest that an expert system (knowledge-based system) will be a powerful tool for the computer-aided design of catalysts.