FUNCTIONAL MODELS IN PERSPECTIVE - THEIR CHARACTERISTICS AND INTEGRATION IN MULTIPLE MODEL-BASED DIAGNOSIS

We analyze functional models in the wide spectrum of model types. Our premise is that, in order to identify the merits of model types, their perceptions of reality are important. We classify modeling approaches into two categories, the kernel and the interpretative. The kernel approach, which assumes that the world can be modeled by a composition of predefined primitives, suffers from important drawbacks. The interpretative approach, to which the functional models belong, assumes that we know how things work. We investigate the characteristics of this approach from a diagnostic point of view and show that, although it provides a powerful tool for describing devices, it is hindered by serious disadvantages when used in isolation. We advertise the use of multiple models as a solution and provide an ontological semantic network as a framework for the integration of multiple model types.