Author(s)

Xiang Shen, Fei Xu, Long Xu, Jidong Cai

Corresponding Author

Xiang Shen

Abstract

Transformers are essential electrical equipment in the power system, and their operating state directly determines the safety, stability, and economy of the power grid. In view of the difficulty in extracting transformer voiceprint features in complex noise environments, a transformer voiceprint feature extraction method based on Variational Mode Extraction (VME) is proposed in this paper. Firstly, the central frequency of the inherent mode component is set according to the mechanism of transformer radiation noise generation, to eliminate the uncertainty caused by the random distribution frequency search method on the decomposition results. Then, with the inherent mode component frequency energy aggregation and the residual signal central frequency energy minimum as the optimization objectives, the cyclic iteration is used to complete the recognition and extraction of transformer voiceprint features, reducing the influence of environmental noise and other equipment noise. The analysis results of on-site actual signals show that this method can effectively reduce the influence of environmental noise and extract clearer and more accurate transformer voiceprint features.

Keywords

Transformer, Voiceprint, Variational mode extraction, Inherent mode component