Qualitative Modelling of Linear Networks in ECAD Applications

Lee M.H., Qualitative Modelling of Linear Networks in ECAD Applications, Expert Update, Vol. 3, Num. 2, pp23-32, BCS SGES, Summer 2000. Qualitative modeling of linear networks in ecad applications (1999) by M Lee Venue: Pages 146?152 of: Proceedings 13th international workshop on qualitative reasoning, QR ?99

There is a real need for qualitative circuit analysis tools in the ECAD application area. Recent simulators based on finite many valued resistance and current models (switch-level models) have been shown to have serious limitations. Through independent parallel research the QR community has produced qualitative electrical models and simulators that have been successfully used in software tools for specific engineering tasks. Many of these systems are based on steady-state models using three valued resistance networks. This paper examines the relationship between qualitative and more conventional electrical circuit modelling and explores the characteristics of certain qualitative formulations with a view to satisfying the needs of future intermediate level models. An existing three valued resistance model is analysed and then extended to deal with additional cases. The problems associated with bridge circuits may pose a major barrier to future progress and this is examined in detail and some solutions are described. The Importance of Qualitative Concepts in ECAD There has been a long standing interest in the ECAD community to find an intermediate modelling level between the gate-level simulators and other analysis tools used in digital logic and the electrical analogue simulators that work at the transistor level. Circuit analysis often involves predictions of both logical/state information and details of electrical parameters and ideally this should be performed in an integrated and coherent environment. While it is possible to model a complete circuit on an analogue simulator this does not capture the distinct qualitative states that characterise important abstractions valued by engineers. Even more important is the problem of maintaining close affinities between the properties of the model and the preferred notations and concepts used by human engineers. This is very difficult to achieve with the voluminous numeric output generated by conventional simulators. This problem is confirmed by reports from ECAD experts who have access to considerable data on engineers ' experience, e.g.

Non peer reviewed