Single-Phase Grounding Fault Location of Distribution Network Based on Physical Information Guidance

Tackling extensive data requirements and the complexity of mechanisms in data-driven fault location for distribution networks, the paper introduces an information-physics-based approach for single-phase grounding fault location in active distribution networks. Firstly, the virtual zero-sequence current injection state of each node is examined to clarify electrical support for physical information guidance. Secondly, the distinct physical properties of power systems are leveraged to supply data for the neural network, enhancing training transparency and robustness. Thirdly, an application framework based on the principle of “counter” is constructed, deriving fault prediction from diverse temporal data sets to enhance precision. Finally, a 10kV network model is simulated in MATLAB/Simulink to test and validate the method, showing it requires some node information and a small number of training sample data to determine the fault line segment.