A State Diagnosis Method for Distribution Transformer Device Based on Cloud Edge Collaboration

A distribution transformer device status diagnosis method based on cloud edge collaboration is proposed to address the difficulty and time-consuming issue of device status diagnosis in new power systems. Firstly, a distribution network device status diagnosis system was designed based on cloud edge collaboration architecture, which improves diagnostic efficiency through data preprocessing in the vicinity. Then, principal component analysis is used on the edge side to extract the main parameters of the distribution transformer, reducing data redundancy. Finally, the Softmax objective function in the XGBoost model was optimized using Focal loss and deployed in the cloud center to analyze the main parameter samples of distribution transformers, thereby obtaining transformer status diagnosis results. Based on the Pycharm platform, experimental analysis was conducted on the proposed method, and the results showed that its diagnostic accuracy reached 96.13%, and the diagnostic time was 12.51s, providing strong guarantees for the safe and stable operation of distribution transformers.