Classification and Detection of Partial Discharge in Cast-Resin Transformers

Abstract

The cast-resin transformers have voltages of up to 72.5 kV with a power rating of up to 60 MVA. Partial discharge (PD) considered as a main cause to the degradation and damage of insulation systems in electrical assets. PD can occur in gaseous, liquid and solid insulating mediums used in assets that are subjected to high electrical fields. This work presents a pattern identification system (PDS), which can effectively and correctly identify the PD pattern of the cast-resin transformer with normal and insulating defects. The PDS combined the clustering scheme with decision tree to diagnose the abnormal defects. The capacity rating of cast-resin transformer is 60 MVA, and the nominal voltage is 22.8 kV. Test results of PDS are compared with two recently developed and publicly available software packages, See5.0 and Weka. The PDS performs much better than the Weka and See5.0 in the aspect of the classification error rates.