Scholarlink

With the high proportion of low-voltage distributed PV access, accurate prediction of distributed rooftop photovoltaic output and user load power is the basis for distribution network operation optimization. However, existing prediction methods cannot meet users’ increasing awareness of data privacy protection and fully reflect the characteristics. Use the user-side meter historical data as input data, this paper propose a photovoltaic output and user load prediction method based on Fully Connected Neural Network and protecting user privacy based on Federated Learning. Taking the Australian public data set as an example for analysis, the effectiveness and accuracy of the proposed modeling method are verified.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

PV Output and User Load Prediction Method in Distribution Networks Based on Federated Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yanlun Li,
  • Xiaobo Dou,
  • Kexin Zhang,
  • Qiangsheng Bo,
  • Pengpeng Lv,
  • Jingwen Yu,
  • Ruipeng Dai

摘要

With the high proportion of low-voltage distributed PV access, accurate prediction of distributed rooftop photovoltaic output and user load power is the basis for distribution network operation optimization. However, existing prediction methods cannot meet users’ increasing awareness of data privacy protection and fully reflect the characteristics. Use the user-side meter historical data as input data, this paper propose a photovoltaic output and user load prediction method based on Fully Connected Neural Network and protecting user privacy based on Federated Learning. Taking the Australian public data set as an example for analysis, the effectiveness and accuracy of the proposed modeling method are verified.