Fault diagnosis of intelligent distribution system based on privacy-enhanced federated learning

In practical applications, different power companies are unwilling to share personal transformer data with each other due to data privacy. Faced with such a data isolation scenario, the centralized learning method is difficult to be used to solve the problem of transformer fault diagnosis. In recent years, the emergence of federated learning (FL) has provided a secure and distributed learning framework. However, the unbalanced data from multiple participants may reduce the overall performance of FL, while an untrusted central server will threaten the data privacy and security of clients. Thus, a fault diagnosis of intelligent distribution system method based on privacy-enhanced FL is proposed. Firstly, a globally shared dataset is established to effectively alleviate the impact of unbalanced data on the performance of the FedAvg algorithm. Then, Gaussian random noise is introduced during the parameter uploading process to further reduce the risk of data privacy leakage. Finally, the effectiveness and superiority of the proposed method are verified through extensive experiments. © 2024 Institute of Scientific and Technical Information of China. All rights reserved.