基于MLP-RF组合模型的配电网高阻接地故障检测方法OA

High-impedance fault(HIF)is difficult to be accurately detected by traditional protection devices due to its low-magnitude fault current and inconspicuous waveform characteristics,and it has become a critical hidden danger in the operational safety of distribution networks.In allusion to the detection problem of HIF in medium-voltage distribution systems,a method of distribution networks HIF detection for based on combined model multi-layer perceptron(MLP)-random forest(RF)is proposed.An HIF equivalent model in 10 kV medium-voltage distribution networks is established,and zero-sequence current data under HIF and normal operating conditions are collected.The zero-sequence current data from two cycles after the fault occurrence are decomposed by means of variational mode decomposition(VMD),thereby extracting multi-scale time-frequency domain features to construct a high-quality feature dataset.The MLP-RF hybrid model is introduced to fully leverage the complementary strengths of the two methods.The simulation results demonstrate that the proposed method can realize high-precision fault identification across various operational scenarios and maintains strong robustness and generalization capability even under noise interference.