CatIO - A Framework for Model-Based Diagnosis of Cyber-Physical Systems

Abstract

Diagnosing cyber-physical systems is often a challenge due to the complex interactions between its individual cyber and physical components. With CatIO (From ‘Causarum Cognitio’, Latin for “(seek) knowledge of causes”), we propose a framework that supports a designer in developing corresponding diagnostic solutions that utilize either abductive or consistency-based diagnosis for detecting and localizing faults at runtime. Employing an interface to tools of the modeling language Modelica, a designer is able to simulate a cyber-physical system’s detailed behavior, and based on the observed data she can then assesses the diagnostic solution(s) under development and explore the trade-offs of individual solutions. For the abductive reasoning variant, CatIO supports also in coming up with the required abductive diagnosis model via an automated concept based on fault injection and the simulation of corresonding Modelica models.