Frequency domain analysis of acoustic emission signals for classification of partial discharges

Online monitoring of partial discharges (PD) can reduce the risk of catastrophic failure of power apparatus. Acoustic emission technique is advantageous for online PD measurement and is adopted here. Partial discharges are classified according to the origin of discharges and their effects on insulation deterioration are also different. This paper reports experimental results and analysis for classification of partial discharges using acoustic emission signal. Acoustic emission signal produced by PD are detected by sensors mounted on the experimental tank surface, stored on an oscilloscope and fed to computer for analysis. The AE signals are analyzed in frequency domain for classification. Amplitude spectrum and power spectrum of acoustic signals are examined. The frequency domain parameters calculated are frequency at peak magnitude, median frequency & percentage signal energy in frequency band. These parameters are used for classification of partial discharges.