(A) method of knowledge base verification and validation for nuclear power plants expert systems

Abstract

The adoption of expert systems mainly as operator supporting systems is becoming increasingly popular as the control algorithms of system become more and more sophisticated and complicated. As a result of this popularity, a large number of expert systems are developed. The nature of expert systems, however, requires that they be verified and validated carefully and that detailed methodologies for their development be devised. Therefore, it is widely noted that assuring the reliability of expert systems is very important, especially in nuclear industry, and it is also recognized that the process of verification and validation is an essential part of reliability assurance for these systems. Research and practices have produced numerous methods for expert system verification and validation (V & V) that suggest traditional software and system approaches to V & V. However, many approaches and methods for expert system V & V are partial, unreliable, and not uniform. The purpose of this paper is to present a new approach to expert system V & V, based on Petri nets, providing a uniform model. We devise and suggest an automated tool, called COKEP (Checker Of Knowledge base using Extended Petri net), for checking incorrectness, inconsistency, and incompleteness in a knowledge base. We also suggest heuristic analysis for validation process to show that the reasoning path is correct.