Partial vs. Corona Discharges in XLPE-Covered Conductors: High-Resolution Antenna Dataset for ML Applications

Dataset provides measurements for the differentiation between partial discharges and corona discharges in XLPE-covered conductors utilized in medium-voltage overhead power distribution lines. Using a contactless antenna-based approach, we collected signals with a sampling rate of \(10^7\) data points over a 20 ms window. Measurements were acquired using two distinct antennas to capture variations in signal characteristics. The dataset spans two days and includes 100 measurements for each discharge type (5 fault classes and 2 background types), alongside background samples recorded with and without high voltage. The acquisition method follows prior research on antenna-based fault detection in covered conductors but specifically addresses the challenge of distinguishing between these two phenomena. Successful classification can enhance diagnostic accuracy in power distribution networks, mitigating risks associated with insulation degradation and potential accidents. By providing a controlled yet realistic dataset, this work enables the development and validation of machine learning models for non-invasive discharge identification. The dataset contributes to improving the reliability of covered conductors, particularly in regions where overhead line safety is a priority.