QUALITATIVE REASONING ABOUT PHYSICAL SYSTEMS - AN ARTIFICIAL-INTELLIGENCE PERSPECTIVE

Some interdisciplinary issues concerning artificial intelligence (AI) are explored in relation to modelling in physics and engineering. A short survey is given of automated qualitative reasoning about physical systems, which in recent years has become an active research area in AI, and has been partly influenced by system dynamics. The conceptual aspects of the AI approach to physical science are critically assessed, and the mutual relevance of the two disciplines is pointed out. In particular, we indicate how bond graphs can be interpreted as an AI knowledge representation language. On this basis, a knowledge-based approach, called qualitative bond analysis (QuBA), is discussed that exploits bond graphs such that a computational AI system is able to deduce qualitative information about physical systems.