Knowledge coding methods for rule-based expert systems

摘要

This paper gives an overview of knowledge representation methods that are currently being implemented for use in a hybrid expert system shell that has been under development at the Department of Control and Instrumentation, BUT. Two approaches are discussed - a diagnostic and a planning expert system knowledge base coding. A new method suitable for diagnostic rule-based expert systems with automated weight tuning, deriving from neural networks, is proposed. Next, an approach to modeling knowledge bases using Petri Nets is discussed and inference engine operations are compared for both diagnostic and planning expert system.