Abstract: Arc high-impedance grounding faults represent the most prevalent fault type in medium- and low-voltage distribution networks. The large transition resistance during initial fault stages and weak fault signals necessitate the development of specialized identification methods to enhance detection effectiveness and accuracy. This paper proposes an arc high-impedance initial fault detection method based on zero-sequence current nonlinear distortion characteristics. Using a constructed 10 kV arc high-impedance grounding fault test platform to acquire zero-sequence current data, we implement segmented filtering and processing while introducing a quantitative representation method for waveform nonlinear distortion. Three distinctive fault characteristics are established based on waveform distortion analysis, enabling arc high-impedance grounding fault identification through predefined criteria. Comparative evaluations with wavelet analysis, current over-limit time method, and harmonic method demonstrate this method's unique capability to identify faults with transition resistances up to 80 kΩ, proving effective for both early-stage arc high-resistance grounding faults and extremely high transition resistance conditions. These findings hold significant theoretical importance and practical value for distribution network initial fault identification and protection systems.