Machine Learning and Digital Twins for RUL Prediction of DC Semiconductor Circuit Breakers

Direct current (DC) Semiconductor Circuit Breakers (SCCBs) are considered as enablers for the further integration of DC systems. Although the reliability of these devices is of crucial importance, conventional testing and lifetime prediction lack the consideration of operating conditions in field application and realtime remaining useful life (RUL) prediction. Within this paper a new approach employing a digital twin enabling digital services for degradation indicator-based RUL prediction using machine learning (ML) is presented and results of a base model implementation for RUL prediction are discussed. In addition, the concept for a novel setup for testing the new services with real world mission profiles is presented.