Meta-diagnosis for a special class of cyber-physical systems: The avionics test benches

Abstract

Several researches have been made to determine the best system representation for process diagnosis. Many approaches were considered, such as heuristic searches, event driven models, structural models... As of today, there are no clear winners considering real applications on industrial systems. In this article, based on a specific system model, a diagnostic approach for complex systems is discussed. The proposed diagnostic algorithm processes a multi-layered system model and converges to the right representation granularity, suitable to explain the fault and to identify the faulty subsystem or component. The method is adapted for a very specific application: the ATB of the NH90, a medium weight multi-role military helicopter. The problem is reformulated as a meta-diagnosis problem since the test bench may also host the fault. A detailed case study is presented for illustration.