Research on the Fault Line Selection Method for Small Current Grounding Systems Based on Time-Frequency Features and Improved CNN

Aiming at the problems of low recognition rate of fault signal features, easy interference of line selection criteria by other fault conditions, and low line selection accuracy in the grounding fault of the small current system, a fault line selection method based on time-frequency features and improved CNN is proposed. First, the improved matrix pencil is used to extract the frequency domain features of the electrical parameters of the fault line, and the feature indicators are constructed together with the time domain feature indicators as the basis for fault line selection. Second, BiLSTM is embedded before the fully connected layer of CNN to enhance its ability to identify feature indicators, and a grounding fault line selection model is constructed. Finally, based on the simulation under different fault conditions, a sample set is constructed, and the proposed feature indicators and line selection model are used to identify the fault line of the system to verify the feasibility and robustness of the proposed line selection method.