Research on the Fault Mechanism and Line Selection of Power Distribution Network with Multi Power Electronic Converters

For single-phase ground faults in power distribution network with DG (Distributed Generation), the distribution of fault currents is significantly affected by DG output regulation, while the time-frequency coupled transient characteristics of zero-sequence currents at the line end are unaffected by DG. Based on this, an intelligent fault line selection algorithm is proposed, which utilizes the differences in the three-dimensional features of zero-sequence current amplitude, phase and energy spectrum. VMD (Variational Modal Decomposition) is employed with modal component frequency band orthogonality constraints and energy density peak synchronization optimization to achieve transient noise suppression and high-purity fault feature extraction. Combined with the IBKA (Improved Black Kite Algorithm) to optimize the center distribution of the RBF neural network’s basis functions, a dynamic feature perception model is constructed. By calculating the energy entropy of the fault line through VMD decomposition and normalizing it before inputting it into the network, the output layer is mapped to a fault line encoding vector. Simulation verification demonstrates its robustness and rapidity advantages in fault line selection under high-noise environments.