Fault line selection for small current grounding system based on ICEEMDAN and improved VGG

In order to solve the problem of low accuracy of single-phase grounding fault line selection method in traditional low-current grounding system, this paper proposes a single-phase grounding fault line selection method for small-current grounding distribution network based on improved VGG. Firstly, the original signal is transformed into an Intrinsic Mode Function component by using the ICEEMDAN algorithm, and then the entropy value of each IMF component is calculated by using the distributed entropy algorithm and differentiated processing, and the processed IMF component is reconstructed to highlight the frequency domain characteristics and amplitude characteristics of the original data, and the influence of noise is removed to a certain extent. The processed zero-sequence current data is converted into a two-dimensional image by the Gramian Angular Field algorithm, and on the basis of retaining the data features, a single-phase grounding fault dataset is prepared, and the fault characteristics are more prominent, which is convenient for neural network processing. Then, the line selection results were obtained by improving the training iterative optimization parameters. Finally, an experimental platform was built by using Simulink, and the proposed algorithm was compared with two line selection methods. Experimental results show that the proposed algorithm has higher line selection accuracy and better noise immunity.