Qualitative monitoring of time-varying physical systems

The authors present an approach to monitoring physical systems which combines quantitative and qualitative techniques. A description of how the technique works is presented, and a prototype software system to perform this monitoring function is described. The qualitative modeling method used relies on the notion of landmark points, which are certain specific values of the system's variables: landmark points are selected in such a way that they delimit regions of qualitatively different behaviors of the system. A qualitative monitoring approach using critical hypersurfaces is described. An example of deriving a critical hypersurface from the knowledge of one landmark point is shown. The monitoring algorithm presented is implemented as a module called Model Selector in the software system COPER/IC.<
>