Improved fault cause identification for overhead distribution grid and field data evaluation

To improve the reliability of our distribution network, we launched fault cause identification in June 2024 as a new function of our advanced distribution automation system (ADAS). By identifying the fault cause before patrols are initiated, the patrol areas can be reduced and the outage duration shortened. To determine the fault cause categories, we analyzed fault data accumulated over the past 15 years and extracted seven common fault categories, including insulators and bird contacts. The estimation process combined statistical analysis using associated information such as the time and weather when the fault occurred, and waveform analysis based on measured zero-sequence current and voltage (I0/V0). To improve the accuracy, coefficients calculated from the V0 amplitudes were utilized in the waveform analysis and a method was developed to extract fault data with associated information similar to that of the target fault. We evaluated the identification accuracy using 89 items of ground fault data collected on our distribution network from March 2022 to April 2024 and found that the above two measures improved the accuracy rate from 49% to 71%.