In order to solve the difficulties in identifying, selecting and locating grounding faults caused by the phenomena of high resistance faults, intermittent grounding faults, arc instability, complex structure of distribution network, complex branch lines and random load distribution, the non-synchronous sampling harmonic measurement algorithm based on TDFT is adopted. The algorithm of current grounding judgement is used to distinguish, locate, and isolate grounding faults. The terminal DTU of station adopts ADSP-BF607 as the main processing chip in hardware. It has dual processing architecture of DSP and ARM, and owns the advantages of fast processing fault data, high accuracy and good recording performance. Based on the TDFT non synchronous sampling harmonic measurement algorithm, the first half wave small current grounding decision method is obtained. In order to effectively judge the inrush current at the moment of switch closing, the DTU adopts the results of Discrete Fourier Transform (DFT) and restrains the spectrum leakage error by weighted algorithm transformation. The traditional algorithm, windowing algorithm and TDFT asynchronous sampling harmonic measurement algorithm are compared and analyzed. The experimental results show that the design of inrush current and small current grounding fault algorithm based on TDFT asynchronous sampling harmonic measurement is accurate and reliable in the aspects of small current grounding and isolation grounding fault of 10 kV distribution network system. This work is supported by Science and Technology Project of State Grid Beijing Electric Power Company (No. 52020116000F) and National Natural Science Foundation of China (No. 51477173).