Fault Location Method of Distribution Network Based on Phasor Feature

A fault location method of distribution network based on stacked denoised auto-encoder (SDAE) is proposed. The method integrates the practicability of single-end ranging and the nonlinear fitting characteristics of artificial intelligence to achieve an accurate range of distribution network lines when faults occur. The improved topology based on actual distribution lines verifies the validity of the SDAE model. The hardware closed-loop verification is carried out at the end of the study. Simulation results show that SDAE is superior to other machine learning methods in fault diagnosis and has high fault location accuracy. In addition, this method is robust to measurement noise and data loss error, which can provide an auxiliary decision for existing fault location methods.