24 September 2021 in IEEE Access
MRMohamad Nur Khairul Hafizi Rohani
Mohamad Nur Khairul Hafizi Rohani
Partial discharge (PD) signal classification analysis on cross-linked polyethylene (XLPE) cables is complex, requiring a comprehensive understanding of the characteristics of PD patterns. In the realm of high-voltage electrical insulation, PD pattern characteristics, such as PD charge and inception voltage, are essential as assessment criteria in diagnostics systems using PD classifiers. This paper provides a review of the various PD patterns and classifiers used by previous researchers, specifically for XLPE cables. In addition, the differences of the studies on various sensor developments based on PD detection in the past 27 years are also discussed. The repeatability, recognition accuracy, recognition speed, and effect of feature sizes on each PD classification method are reviewed and explained. This review indicates that the pattern recognition for PD signal using artificial neural network (ANN) exhibits better performance than the other methods in terms of accuracy and repeatability, and the reduction of feature size does not affect the accuracy of ANN.
21 July 2026 in SSRN Electronic Journal
Reliable detection and classification of partial discharge (PD) activity are essential for assessing the condition of solid insulation systems, yet measurement variability and feature redundancy can reduce diagnostic reliability. This study presents an experimental investigation of PD behaviour in epoxy resin insulation under four controlled defect conditions, namely air-filled voids, water-filled voids, metallic particle inclusions, and surface discharges, using a broadband Rogowski coil sensor under IEC 60270-based measurement procedures. Signal quality was evaluated through acquisition consistency, repeatability, and waveform stability over 1000 consecutive AC cycles, where the coefficient of variation of the RMS signal remained below 5% for all PD types. Phase-resolved partial discharge (PRPD) patterns were constructed and represented using six phase-resolved distributions, from which 30 statistical features were extracted. Principal component analysis (PCA) was then applied to reduce feature redundancy, and the optimized representation retained 6 principal components while preserving more than 95% of the cumulative variance. The reduced feature space was further validated through inter-dataset stability and class separability analysis across five independent datasets. A genetic algorithm-optimized backpropagation neural network (GA-BPNN) was subsequently implemented for multi-class PD classification. The proposed framework achieved a classification accuracy of 95.0%, demonstrating that integrating experimental signal-quality validation with optimized PCA-based feature reduction improves the reliability, compactness, and effectiveness of PD measurement and diagnosis in epoxy resin insulation systems.
21 July 2026 in SSRN Electronic Journal
Reliable detection and classification of partial discharge (PD) activity are essential for assessing the condition of solid insulation systems, yet measurement variability and feature redundancy can reduce diagnostic reliability. This study presents an experimental investigation of PD behaviour in epoxy resin insulation under four controlled defect conditions, namely air-filled voids, water-filled voids, metallic particle inclusions, and surface discharges, using a broadband Rogowski coil sensor under IEC 60270-based measurement procedures. Signal quality was evaluated through acquisition consistency, repeatability, and waveform stability over 1000 consecutive AC cycles, where the coefficient of variation of the RMS signal remained below 5% for all PD types. Phase-resolved partial discharge (PRPD) patterns were constructed and represented using six phase-resolved distributions, from which 30 statistical features were extracted. Principal component analysis (PCA) was then applied to reduce feature redundancy, and the optimized representation retained 6 principal components while preserving more than 95% of the cumulative variance. The reduced feature space was further validated through inter-dataset stability and class separability analysis across five independent datasets. A genetic algorithm-optimized backpropagation neural network (GA-BPNN) was subsequently implemented for multi-class PD classification. The proposed framework achieved a classification accuracy of 95.0%, demonstrating that integrating experimental signal-quality validation with optimized PCA-based feature reduction improves the reliability, compactness, and effectiveness of PD measurement and diagnosis in epoxy resin insulation systems.