An improved high-impedance fault identification scheme for distribution networks based on kernel extreme learning machine
Wanxing Sheng, Keyan Liu, Dongli Jia +1 more
In distribution networks, high-impedance faults (HIFs) occur frequently and have a harmful impact on the distribution network. However, fault detection and fault phase selection of HIFs are challenging due to weak fault characteristics. Therefore, this paper proposes an improved HIF identification scheme based on a kernel extreme learning machine (KELM) that can sensitively identify HIFs and select the fault phase by adaptively extracting the weak fault characteristics. First, a fault feature extraction strategy based on discrete wavelet decomposition (DWT) and the Hilbert–Huang transform (HHT) is proposed to obtain multiple features that describe the weak fault characteristics of HIFs. Second, an XGBoost-based fault feature selection scheme is proposed for screening with sensitive characterization of HIFs. Next, a sensitive and accurate HIF identification scheme based on the improved…