Partial discharge signal denoising using wavelet techniques-on site measurements

Abstract

Partial Discharge (PD) diagnosis is the recognized method to detect defects in the insulation of power system components. However PD measurements are mostly affected by electromagnetic interferences. Wavelet is an effective tool in extracting the PD signal from noise. In this paper a new wavelet threshold determination method called multi-level wavelet decomposition is proposed to denoise the PD signal. With the implementation of this method the PD measurement sensitivity is increased. The effectiveness of this is shown by comparing signal to noise ratio (SNR) with FFT and Discrete Wavelet Transform (DWT).