When tree-related high-impedance fault (THIF) occurs in distribution network, the fault characteristics change weakly and the fault state changes randomly, which makes it difficult for existing fault detection methods to identify the fault effectively. In order to realize the effective detection of THIF, a 10kV overhead line single-phase contact tree ground fault detection method based on multi-resolution wavelet transform and the multi-parameter information fusion is proposed. Firstly, the zero-sequence current is decomposed by multi-resolution wavelets to obtain eight frequency bands of signal, followed by standard deviation, second-order center distance, interquartile difference, and the mean value as the screening index to screen out the target frequency band, and reconstruct the signal in the target frequency band, and finally the power spectral density analysis(PSD) of the reconstructed signal is carried out to calculate the number of peaks, and the threshold is preset for comparison, and the fault detection is realized by using the number of overruns as the fault criterion. After a lot of field tests and algorithm comparison, the method can identify the fault accurately, is not affected by the accuracy of measuring equipment, and can adapt to the change of fault state effectively.