CPS-based fault-tolerance method for smart factories

摘要

In this paper, we propose a fault tolerance method in smart factories which features a simulation-based framework and a fault management application, FORTUNA (Fault-tOleRanT and aUtomated decisioN making Application). The simulation-based framework is characterized by a Virtual Factory (VF) which plays an integrator role in CPS architecture for the smart factory. The VF provides two key functionalities, (1) a simulation of CPS models represented by ECML (ETRI CPS Modeling Language), and (2) a state estimation based on massive CPS data. By deploying Kalman filter, a predictor-corrector type estimator, FORTUNA not only detects faults through the realtime observation of CPS data but also predicts potential failures by the simulation. It is demonstrated that the method can work effectively in industrial systems with real dataset.