Partial Discharge Pulse Clustering Analysis using Wavelet Decomposition in Power Cables

The wavelet decomposition technique was studied to develop a more improved noise removal technique in partial discharge diagnosis. For the purpose, DC partial discharge pulse waveforms measured at 5m,25m, and 75m positions of the power cable were decomposed into sub-waveforms of high-frequency and low-frequency components. The sub-waveforms of the high frequency component were well classified in the frequency-related characteristics and those of the low frequency component were well classified in the magnitude-related characteristics. The PCA clustering technique to the features extracted from the sub-waveforms showed a very good classification result.