Opportunistic multimodel-based diagnosis. Framing all the knowledge we have to diagnose complex artifacts

Second-generation expert systems try to reduce the dependency from experts for knowledge acquisition working on models of the domain of application, instead of models of expert's experience. In industrial applications, it is possible to elicit different types of possibly incomplete and approximate models. We propose a framework to organize domain knowledge about artifacts in models defined along three dimensions: level of abstraction, uncertainty, and epistemological type (teleology, process, function, structure). All the knowledge is represented using possibly qualitative relationships among variables and possibly approximated values. OMISSYS (Opportunistic Multi-model-based diagnosIS SYStem) uses the so-framed incomplete, uncertain, and approximate knowledge to diagnose a complex artifact, such as an industrial plant. It navigates opportunistically through an arbitrary set of models to reach a predefined diagnosis goal.