Analysis on vibration and acoustic joint mechanical fault diagnosis of high voltage vacuum circuit based on wavelet packet energy relative entropy

In view of the requirement of high reliability and small disturbance in the condition of high-voltage (HV) vacuum circuit breaker (VCB), this paper puts forward a method of fault diagnosis based on online monitoring of vibration and acoustic signal. The research object is the 12kV indoor high-voltage vacuum circuit breakers and the system of fault diagnosis of the research object is built. This method uses fast kernel independent component analysis to make blind source separation processing for acoustic signals collected when HV VCB switch on. The wavelet packet energy should be calculated in each frequency of the vibration and acoustic signal. Support vector machine with the wavelet packet energy relative entropy as input vector to classify the common states, such as the normal condition, insufficient lubrication of the crank arm and mechanism fall off in moving. Experimental study has shown that, according to research on the characteristics of the vibration and acoustic signal generated by the circuit breaker, the proposed method can differentiate between normal working state and failure state and meet the high reliability of HV VCB status monitoring requirements.