Abstract: Aiming at the problems of low efficiency and error-prone manual recording in type tests of distribution network grounding fault automatic compensation devices, an automatic test recording program is designed using AI-assisted programming. The program is developed based on Python and Flask framework, which can parse COMTRADE files in IEEE Std C37.111 standard, calculate fundamental RMS through DFT algorithm, identify fault and compensation moments by 3U0 mutation, and automatically complete data extraction, parameter calculation and test record generation. Experimental results show that the recognition error of the program is less than 10 ms, which can significantly improve test efficiency and data accuracy, and meet the type test requirements of DL/T 1057-2023.