Abstract:

Aiming at the problems of difficult fault line selection and slow detection for the small and medium current grounded systems in distribution networks,a multi criterion fusion method for line selection of small current single-phase ground fault based on convolutional neural networks (CNN),Fourier transform and wavelet packet decomposition is proposed.CNN is applied to the problem of line selection for small current single-phase ground fault.Some components of the zero-sequence current of each line after processing are taken as input, and the fault line is taken as output.After a large amount of data training, the mathematical model of the fault line is obtained.When the new fault data is input into the model,the fault lines can be selected based on the mapping relationship.Finally, the feasibility of the proposed method is verified by the simulation using MATLAB/Simulink.

Key words: fault line selection, small current grounding, convolutional neural network (CNN), multi criterion fusion