Pattern Recognition for UHF Partial Discharge of Power Transformer Using LS-SVM
Partial discharge diagnosis is an important tool for detecting insulation defects in power equipments. This paper presents a pattern recognition approach based on Least Squares Support Vector Machine (LS-SVM) for Ultra High Frequency (UHF) partial discharge diagnosis of power transformer. Six different feature parameters were extracted from the data obtained from Partial Discharge (PD) on-line monitoring system. LS-SVM was used to discriminate between 4 different PD sources. Experimental results demonstrate that the proposed approach has higher recognition accuracy compared with traditional BPNN recognition method under condition of small samples, and has great potential for use of field data.
J. S. Yuan and H. K. Shang, "Pattern Recognition for UHF Partial Discharge of Power Transformer Using LS-SVM", Advanced Materials Research, Vols. 588-589, pp. 384-387, 2012