State Identification of Transformers under DC Bias Based on Nonlinear Vibration Feature

Power transformers under DC bias means that there is a DC component in the magnetic flux, which has an important impact on the vibration of transformer. In this paper, the mechanism of the transformer vibration caused by magnetic DC bias is reviewed. Next, a feature extraction method based on vibration mutual information is proposed, and a neutral point current prediction model based on the extreme learning machine (ELM) algorithm is also established. Finally, the experiments are conducted to verify the feature extraction method, and the neutral point current prediction model is trained and tested. The results show that the extracted nonlinear vibration feature combined with the ELM method can identify the transformer state under DC bias.