CATALisp - an expert system for real-time diagnosis of multiple-failure transients in pressurized water reactors

The state of the art in artificial intelligence (AI) and expert system (ES) technology has matured to a degree that the potential development of a computer-aided/automated diagnostic and transient mitigation system in the area of nuclear reactor operation can be considered. Since traditional methods cannot handle complex systems efficiently, AI techniques provide a means to emulate an expert reactor operator rather than follow mechanistic methods. Computer-aided transient analysis coded in LISP (CATA-Lisp) is a confidence level based expert system written in Common LISP on the SYMBOLICS 3640 computer system. New versions are being developed in Common LISP for the Texas Instruments (TI) Explorer and the Sun microsystems machines, CATALisp manipulates both a knowledge base of transient identifier patterns (tree structured to allow for zooming in diagnostics) and a knowledge base containing a qualitative model of a nuclear power plant. The interference engine used by CATALisp uses the information stored in both knowledge bases to arrive at confidence level values that are used to infer particular plant states.