Winding Mechanical Fault Diagnosis Technique of Power Transformer Based on Time-Frequency Vibration Analysis

Abstract

Huge electromagnetic force during sudden short circuit can cause severe winding mechanical fault. Transformer tank vibration characteristics are closely related to winding mechanical condition. Transient vibration response on transformer tank under short circuit impact can be utilized to monitor winding condition and diagnose winding fault. Short circuit impact experiment according to IEC standard is performed on a 110kV transformer. 33 times impacts were performed at A-phase and tank vibration signal is recorded every time during the impact. Since transformer transient vibration signal is non-linear and non-stationary, this paper utilizes Hilbert-Huang Transform (HHT) methodology to analyze transformer vibration signal in time-frequency domain. 200Hz-1000Hz component in Hilbert spectrum is deeply studied. Image processing method is used to quantity the occurrence number of frequency above 200Hz in Hilbert spectrum. The result shows 800Hz-1000Hz component increases largely during 23rd impact compared with the previous impact result. Thus, 800Hz-1000Hz component in Hilbert spectrum can be regarded as an indicator to diagnose winding fault. What's more, Short Circuit Impedance (SCI) experiment shows that SCI variation rate of A-phase changes 1 % after 24th impact and increases to −21.85% after 33rd impact. Suspended core experiment after the short circuit impact experiment reveals that severe winding mechanical fault occurs in both high voltage winding and medium voltage winding. Both of the results validate the correctness of Hilbert spectrum analysis of transient vibration response. Therefore, time-frequency vibration analysis can be used to monitor winding condition and diagnose winding fault.