Separation of Partial Discharge Sources by Application of Feature Extraction and Clustering Methods

Identifying the origin of partial discharges (PD) is one of the most significant challenges in PD measurement and analysis. Conventionally, this task relies on the inspection of phase-resolved partial discharge (PRPD) patterns, which demands considerable expert interpretation and is often complicated by the superposition of signals originating from multiple defect mechanisms and disturbance signals/noise.