Partial discharge monitoring in a noisy environment

Defining the resolution levels where a PD signal is localized and selecting an automated threshold values for on-line de-noising and measurement are the main challenging task in wavelet multi-resolution analysis (WMRA) application for PD detection and measurement. This paper proposes a new wavelet-based technique for monitoring PD signals embedded in high noise levels. The data is decomposed while sliding into Kaiser's window. Only the maximum coefficients extracted at each resolution level are used to extract the PD signal from noise and measure its magnitude. No thresholding or reconstructions of the thresholded coefficients are required. The extracted data-size of PD signal is very small as compared with actual signal. The proposed monitoring technique is applied on simulated data and laboratory data and gave good results. The simulated data are constructed by using real noise collected from laboratory measurements.