A first-principles generic methodology for representing the knowledge base of a process diagnostic expert system

In this paper we present a methodology for identifying faulty component candidates of process malfunctions through basic physical principles of conservation, functional classification of components and information from the process schematics. The basic principles of macroscopic balance of mass, momentum and energy in thermal hydraulic control volumes are applied in a novel approach to incorporate deep knowledge into the knowledge base. Additional deep knowledge is incorporated through the functional classification of process components according to their influence in disturbing the macroscopic balance equations. Information from the process schematics is applied to identify the faulty component candidates after the type of imbalance in the control volumes is matched against the functional classification of the components. Except for the information from the process schematics, this approach is completely general and independent of the process under consideration. The use of basic first-principles, which are physically correct, and the process-independent architecture of the diagnosis procedure allow for the verification and validation of the system. A prototype process diagnosis expert system is developed and a test problem is presented to identify faulty component candidates in the presence of a single failure in a hypothetical balance of plant of a liquid metal nuclear reactor plant