| Title |
PD Classification Method for Eco-friendly Gas-Insulated Power Equipment |
| Authors |
홍성준(Seong-Joon Hong) ; 손진근(Jin-Geun Shon) |
| DOI |
https://doi.org/10.5370/KIEE.2026.75.9.2320 |
| Keywords |
Partial Discharge; Eco-frienldy Gas-Insulated Power Equipment; Neural Network; Backpropagation Algorithm |
| Abstract |
This paper investigates partial discharge(PD) classification in eco-friendly gas-insulated power equipment using g³. Four representative PD defects were simulated, and nine time-domain, frequency-domain, and statistical parameters were extracted from UHF signals. Artificial neural network(ANN), support vector machine(SVM), and k-nearest neighbor(KNN) algorithms were evaluated using 1,000 PD pulses with five-fold cross-validation. The ANN showed the highest average accuracy of 92.7%, followed by the SVM and KNN. These results demonstrate the feasibility of machine learning-based PD defect classification in eco-friendly gas-insulated power equipment. |