Abstract:
In view of the problem of low accuracy of amplitude comparing line selection method based on traditional fifth harmonic, a single-phase fault line selection method of power distribution network based on fifth harmonics energy and LM-Elman neural network was proposed. Firstly, wavelet packet was used for three-layer decomposition and reconstruction of fifth harmonics in zero-sequence current, and the total energy of the third layer reconstructed wavelet coefficients was obtained; then normalized energy values were used as input of Elman neural network, LM algorithm was used for training and testing. The simulation results show that the method can accurately select fault line in distribution network.