Fault diagnosis for high voltage circuit breaker based on VMD energy entropy and support vector machine

A new fault diagnosis method based on variational mode decomposition (VMD) is proposed in order to realize the fault identification of high voltage switches. Firstly, after preprocessing vibration signal, the feature extraction based on VMD energy entropy is analyzed. Subsequently, SVM is utilized to calculate fault type by utilizing the extracted features. Parameter optimization is solved by grid search algorithm to improve the performance of SVM. Verification of the scientific nature of this method for fault diagnosis is provided by field data, which was obtained through an experiment on a SF6 circuit breaker with a high 35kV voltage. The break vibration data was measured under normal operation, actuator fault, buffer spring invalidity, and loosing of base screw.