Advanced de-noising of power cable Partial Discharge signals by Empirical Mode Decomposition

As Partial Discharge (PD) signals measured on-site are often contaminated with noise, signal processing techniques are required to obtain optimal PD-detection and PD-location. Additionally, automated processing of measurement data would be desirable to ease condition monitoring of HV equipment in e.g. power cables. While the wavelet thresholding (WT) is a widely utilized method to de-noise such PD signals, the Empirical Mode Decomposition (EMD) could be a beneficial method as well. To compare the de-noising results, localization of the cable PD is done with Time Domain Reflectometry (TDR) by means of a robust energy criterion (EC) and a phase method (PM) algorithm.