Model Synthesis for Model-Based Fault Analysis

Several model-based reasoning tasks benefit from the use of models at multiple levels of abstraction, where a system is described hierarchically as a set of subsystems (which can in turn be decomposed down to basic components). Providing models at multiple level of abstraction by hand is not an option because it would easily lead to inconsistencies. In this paper we propose an approach to synthesize the model of a subsystem starting from the models of its components with the goal of making it usable for a human modeller. In particular, we aim at producing a model which is reasonably compact (thus trying to avoid redundancies) and is expressed in the same intensional format (i.e. the same constraint language) and with the same modelling style as the original low-level models. This approach is thought as part of a software tool for fault analysis and diagnosis. Several tasks could in fact benefit from automated model synthesis, in particular model validation and debugging.