Partial discharge developing stages identification based on cluster-hierarchical decision SVM in oil-paper insulation

Based on rod-plane electrodes in oil-paper insulation, the experiments for developing characteristics of partial discharge (PD) were performed using constant voltage method. Four statistical spectra and 29 characteristic values were extracted on the basis of the phase-resolved partial discharge (PRPD) pattern. Then 8 new parameters were extracted from the 29 characteristic values by using kernel principal component analysis and the different developing stages of PD were divided with system clustering. According to the clustering results, the PD stages hierarchical identification method was established based on clustering-multistage Support Vector Machine. Identification results show that the proposed method can identify discharging stages well and the obtained results are basically consistent with the results of system clustering analysis for the same data.