Integrating empirical and ontological knowledge in an expert system for diagnosis of a steam condenser

Presents a knowledge-based expert system designed to solve a problem of industrial relevance: online fault detection and diagnosis of the steam condenser subsystem in a thermal power plant. The expert system features a composite inference mechanism, which utilizes both empirical and ontological knowledge for generating appropriate diagnosis of an observed malfunction. Ontological knowledge is represented by means of a novel representation language, the component-based language (CBL). The main features of the proposed paradigm include the use of explicit fault models and the consideration of credibility and cost of the measurements that drive the diagnostic process. The proposal has been implemented on a Symbolics 3640 Lisp machine, and has been tested on a simplified model of a steam condenser with encouraging results.<>