An EHV Transformer Fault Diagnosis Method based on Acoustic Perception and Pseudo-Labels Intensive Training-CNN

The importance of power transformers is more prominent in the context of EHV grid interconnection, and their state health is the basis of grid operation stability. The classic ways of monitoring transformer faults all have some impact on the transformer. Consider incorporating acoustic detection without transformer outage detection and placement of intrusive sensors. Meanwhile, in order to solve the problem of insufficient labeling of test data, an EHV transformer fault diagnosis method based on acoustic perception and pseudo labels intensive training-CNN is proposed in combination with semi-supervised training. First, the convolutional neural network is trained to ensure the classification accuracy of the model using deep learning ideas. Then, the pseudo-labels idea is introduced and the intensive training mechanism is proposed to improve the classification accuracy of the model again. Finally, the transformer acoustic data are processed and integrated as a way to validate the model performance. Experiments show that the proposed model can further improve the classification performance of the model by using the pseudo-labels intensive training mechanism on the basis of ensuring the classification accuracy. The final accuracy rate can reach more than 92.4 percent.