Qualitative reasoning and cyber-physical systems: abstraction, modeling, and optimized simulation

Refs: 27International audienceComplex systems modeling and simulation are critical in many industrial and research fields and specifically to predict, prove, verify, and understand the behavior of cyberphysical systems. The diversity of variables in a system creates complexity [1] and a need for more efficient modeling and simulation methods. In the case of hybrid systems [2], the heterogeneity of the discrete and continuous parts makes these tasks more challenging by adding the necessity to manage different types of variables and transitions separately. Qualitative reasoning offers a paradigm to study the behavior of such systems with a high level of abstraction, trading precision and specificity against generality and formalism. This paradigm can be preferred to numerical analysis in specific situations, especially in the upstream study of a system, in its design phases. However, the different representations and the various contexts of such systems create an important obstacle to define a general methodology for applying qualitative reasoning and modeling to every case. In this article, we propose a method and a tool to unify different qualitative reasoning techniques on complex cyber-physical systems. We will also develop the possibilities offered by the obtained abstraction in various tasks such as formal proof, verification, property analysis, diagnosis, simulation driving, and system monitoring