Analysis and Detection of Single-Phase Arc Grounding Fault in Ungrounded System with Distributed Generation

The research theory on single-phase grounding faults in distribution networks is relatively mature. However, with the large-scale integration of distributed generation (DG) into distribution networks, its impact on the transient characteristics of small-current grounding faults needs to be further clarified. In particular, during the occurrence of single-phase arcing grounding faults in distribution networks, the restriking of the arc will further increase the complexity of the transient process. This paper establishes fault models for the integration of inverter-based and rotating distributed generation into distribution networks using the symmetrical component method. Combined with existing black-box arc models, it studies the variation law of the transient process. Then, a single-phase grounding fault simulation model is built on a simulation platform to verify the correctness of the fault analysis. The results show that the arc introduces new transients into the steady-state process of zero-sequence current; the inverter-based distributed generation has little impact; and the connection position of the rotating distributed generation affects the amplitude and frequency of the fault transient current. Finally, a method combining Convolutional Neural Network(CNN) and Long Short-Term Memory(LSTM) network is used to detect single-phase grounding faults, and the results demonstrate the accuracy of the detection method.