Advanced detection of multiple PD sources in cables using time-frequency transformations

This paper presents a novel approach for detecting, identifying, and localizing partial discharge (PD) sources within power cable insulation using advanced time-frequency analysis and discrete wavelet transform (DWT). The proposed method effectively distinguishes multiple PD sources with high precision by integrating discharge energy and time-frequency index (TFI) analysis. Experimental validation across seven scenarios involving different cavity combinations within the cable insulator demonstrates the method's robustness, even in environmental noise. The synergy between discharge energy metrics and TFI enhances diagnostic accuracy by capturing sharp discontinuities in the time-frequency domain. The results confirm the capability of the proposed approach to identify PD sources at various locations, including the cable ends, joints, and along the insulation. PD faults manifest as deformations in the TFI and energy plots within the time-frequency domain. The fault location can be identified based on the cable length and deformation time. This comprehensive framework improves PD source separation and localization and reduces computational costs, making it particularly beneficial for machine learning-based data clustering.