Abductive Diagnosis based on Modelica Models

When employing model-based diagnosis, coming up with the required model is a knowledge and resource-intensive task. In this article, we show how to automatically derive an abductive diagnosis model from system models written in the popular Modelica modeling language. Using Modelica’s simulation features we inject potentially faulty component behavior into the system model and automatically derive cause and effect rules. We afterwards use these rules for performing abductive diagnosis of the malfunctioning system. Our first case studies demonstrate the applicability and viability of the approach.