Feature Extraction of Transient Zero-Sequence Voltage and Adaptive Protection Criterion for High-Resistance Ground Faults in Low-Voltage Distribution Networks

High-resistance ground faults (HRGFs) in low-voltage distribution networks are highly concealed. Traditional zero-sequence current protection and overvoltage protection are prone to failure to operate or maloperation, making it difficult to meet the requirements of safe power supply. This paper aims to improve the sensitivity and adaptability of HRGF detection by optimizing feature extraction methods and constructing adaptive protection criteria through in-depth analysis of transient zero-sequence voltage signals. The study employs signal preprocessing techniques such as wavelet denoising and adaptive filtering, combined with multi-dimensional feature extraction algorithms in the time domain, frequency domain, and time-frequency domain, and designs an adaptive threshold adjustment mechanism based on real-time operating conditions. Through MATLAB/Simulink simulation and laboratory experiments, the proposed method achieves an HRGF detection accuracy of 98.7% within the fault resistance range of 500Ω~5000Ω, with an action time of less than 20ms. This effectively solves the problem of insufficient sensitivity of traditional protection schemes for high-resistance faults and provides reliable technical support for HRGF protection in low-voltage distribution networks.