Abstract: In order to improve the accuracy of automatic detection of insulation faults in distribution network bypass cables, ensure the reliability and safety of power grid operation, a research on automatic detection method of insulation faults in distribution network bypass cables was carried out using improved wavelet entropy. Firstly, develop a plan for collecting electrical parameters of distribution network bypass cables and collect electrical parameters. Secondly, by utilizing improved wavelet entropy from the collected data, the characteristics of cable insulation faults can be effectively extracted. On this basis, the wavelet entropy ratio of two adjacent detection points on the cable is calculated to locate the fault section, and the voltage drop comparison method is used within the section to locate the insulation fault point, achieving the goal of automatic insulation fault detection. The experimental results show that after application, this method can accurately detect insulation fault points in cables, and the error between the detection position and the actual position is controlled within 2 meters, with high detection accuracy and strong robustness.