Prognostics algorithms for circuit breaker application: a benchmark analysis

Today, globalization of markets and more stringent requirements from customers put forward a multitude of challenges to managers. They strive to optimize all systems involved in their organizations in order to outperform on all the main Key Performance Indicators (KPIs). Maintenance as a system plays also a key role in order to reach the KPI needed for equipment reliability and availability. To this end, also maintenance strategies have evolved from standard cyclic maintenance to more advanced approaches like Condition Based Maintenance (CBM) and, as a further evolution, Predictive Maintenance (PdM) approach. Even if lot of work has been done regarding these methodologies applied on machines or industrial plants, due to the need of increasing their efficiency and productivity, the same care is not applied to other critical devices, not strictly part of single machines, like those related to power distribution. In the case of Low Voltage Circuit Breakers (LVCB), in particular for critical applications, conservative approaches are usually applied. That implies a scheduled replacement of devices could occur with a significant Remaining-Useful-Life (RUL). This paper aims to test different approaches to PdM in this class of devices, in order to evaluate, in a practical way, the performance of different types of features extraction algorithms, suitable for the development of a prognostic system, in a real industrial application. More in detail, the paper will compare three types of promising algorithms: a custom Dynamic Time Warping, an Empirical Mode Decomposition combined with Sampled Entropy and a lightweight analysis based on Fast Fourier Transform.