Fault Detection of High Voltage Circuit Breaker Based on Moving Window Variational Mode Decomposition and Fuzzy Entropy

Vibration analysis has been recently proven to be a powerful and reliable treatment for noninvasive mechanical fault diagnosis of high-voltage circuit breakers. The computation intensity makes conventional methods based on the traditional variational mode decomposition (VMD) invalid for the case of real-time scenario. This paper presents a moving window strategy for real-time VMD implementation, which is suitable for decomposition of long signals with high sampling frequency. The similarity of the reconstructed signal in the segments can be calculated by fuzzy entropy through the soft and continuous boundaries of fuzzy functions, which earns stronger relative consistency and less dependence on data length. A case study of three different faults simulated on a real vacuum breaker yields that the results are more efficient than the other traditional methods for the different size of moving window.