Abstract: Ground transition resistance is high, the steady-state characteristics of single-phase ground fault are not obvious, and it is difficult to distinguish it from normal disturbance, and the accuracy of fault identification is greatly affected. To address the above problem, wavelet analysis is used to decompose the fault signal at three levels to obtain different high-frequency components to construct the feature vectors, and dendritic neural networks are used to classify and identify the fault feature vectors. The simulation model is established in MATLAB/Simulink, and the test results show that the method can accurately and quickly classify single-phase grounding high-resistance faults and normal disturbances, and its convergence speed and classification accuracy are better than those of general intelligent algorithms. At the same time, this method still has a high accuracy when the signal is noisy.