Meta-Domains for Automated Model Building

We present a new knowledge representation and reasoning framework for modeling nonlinear dynamical systems. The goals of this framework are to smoothly incorporate varying levels of domain knowledge and to tailor the search space and the reasoning methods accordingly. In particular, we introduce a new structure for automated model building known as a meta-domain which, when instantiated with components, tailors the space of candidate models to the system at hand. The xmission-line meta-domain, for instance, generalizes the notion of an electrical transmission line, using an iterative template to build models it may be customized to specific model-building domains ranging from mechanical vibrations to thermal conduction. and combine this with ideas from generalized physical networks, a met a-level representation of idealized two-terminal elements, and a hierarchy of qualitative and quantitative analysis tools, to produce dynamic modeling domains whose complexity naturally adapt to the amount of available information about the target system.