De-Noising of Partial Discharge Signals in HV XLPE Cables by Reference Noise based on the Wavelet Transform.

The use of underground XLPE cables in power systems is increasing due to the problems associated with overhead lines. However, insulation failure, primarily caused by partial discharge (PD) phenomenon, remains a major concern. Monitoring the insulation condition of XLPE cables, particularly the magnitude and trend of PDs, is crucial. Detecting and localizing PDs online pose a significant challenge due to the presence of various environmental noises that must be eliminated. In recent years, power systemintroduced noises like impulse noises generated by power electronic devices have become prevalent. Therefore, effective de-noising techniques are required to filter out the noises and interferences from the detected PD signals. This paper proposes the use of wavelet transform, which encompasses both time and frequency domains, to remove various noises from PDs, considering their non-stationary nature. To determine the suitable mother wavelet transform, threshold, and number of decompositions, experimental tests are conducted to investigate the characteristics of PD signals in the cables. Additionally, background noise at the measurement site, which varies in substations, is recorded as reference noise for the wavelet-based noise removal process. Using MATLAB software, reference noises and noisy PD signals are measured in substations, and the de-noising process is executed. The proposed method is evaluated on a sample cable, and the results are discussed. Furthermore, the method is applied to detect PD signals in multiple 20 kV substations in Iran, utilizing high-frequency current transformers connected to shield wires, oscilloscopes with high-frequency bandwidth, and MATLAB software. [ABSTRACT FROM AUTHOR]