Transformer Partial Discharge Recognition Based on Phase-Preserving Data Augmentation

The identification of transformer Partial Discharge (PD) is crucial for the safe and stable operation of modern power systems. Traditional PD identification methods rely on manual feature extraction, which has limited generalization capability. Moreover, when applying deep learning algorithms to Phase-Resolved Partial Discharge (PRPD) patterns, conventional data augmentation techniques may disrupt the phase distribution, leading to the loss of critical features and affecting identification accuracy. To address this issue, this paper proposes a PD identification method based on phase-preserving data augmentation techniques. By incorporating Gaussian noise addition, phase shift processing, and segmented perturbation processing, the proposed method effectively retains the phase information in PRPD patterns, avoiding the loss of critical features during data augmentation. Experimental results show that the proposed method significantly outperforms traditional methods in terms of accuracy and stability, achieving a maximum accuracy of 96.39%, demonstrating great potential for practical engineering applications.