Intelligent real-time control of moulding mixture composition in foundries

Abstract

The implementation of an adaptive knowledge-based system in the foundry industry is presented. The problem of moulding mixture composition was solved because the majority of casting defects are due to inappropriate mould quality. The original learning method and system architecture was designed to ensure simple interaction and maintenance of the knowledge base as well as unambiguous interpretation of its content. Elimination of the human factor was achieved with fully automated real-time data acquisition. The method proposed was successfully tested in standard industrial conditions during a period of two years.