Feature Fusion of Pulse Current, Ultrahigh Frequency, and Photon Count Signal: A Novel Discharge Pattern Recognition Method of Metal Particles in GIS/GIL
Xianhao Fan, Hanhua Luo, Fangwei Liang, Tiejun Ma, Jun Hu, Chuanyang Li, Jinliang He
Abstract
The identification of defect-induced discharge patterns can effectively assess the operational risk of gas-insulated electrical equipment. However, accurately identifying the complex patterns of defect-induced discharge behavior based on multisignal remains a challenge. To address this, a partial discharge (PD) pattern recognition method that utilizes multidimensional signals and feature fusion is proposed. In this work, pulse current, ultrahigh frequency (UHF) pulses, and photon counting (PC) signals induced by metal particle defects are recorded. These signals are then used to construct phase-resolved PD (PRPD) and photon counting (PRPC) patterns under increasing voltage, with features extracted to quantify discharge evolution. Finally, a strategy based on dimensionality reduction and clustering analysis is established to develop a recognition model for discharge patterns. In this regard, a novel evaluation strategy for discharge evolution is reported, which is expected to provide new insights for the analysis of defect-induced discharge patterns.