Scholarlink

To address the problem of low forecasting accuracy of the existing methods for short-term power load, this paper proposes a short-term power load forecasting model, which is based on Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Informer. Firstly, a dataset is constructed based on the multi-dimensional power data of a city in Northwest China and the characteristic parameters are analyzed and extracted accordingly. Secondly, the Informer model is involved and then the CNN is used to extract the data features, and the GRU is utilized to capture the extracted data features. Finally, the proposed model is used to accomplish the accurate forecasting of short-term power load. It is shown that the RMSE values of the CNN-GRU-Informer model are only 0.0228 and 0.0258 in the A and B datasets, respectively. Therefore, the proposed model can well achieve the short-term power forecasting and provide a reference for other forecasting methods.

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

Short-term Power Load Forecasting Based on the CNN-GRU-Informer Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Xiaohua Wan,
  • Yong Wei,
  • Hui Fan,
  • Panrun Jin,
  • Xushan Zhao,
  • Xiaoyuan Wei

摘要

To address the problem of low forecasting accuracy of the existing methods for short-term power load, this paper proposes a short-term power load forecasting model, which is based on Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Informer. Firstly, a dataset is constructed based on the multi-dimensional power data of a city in Northwest China and the characteristic parameters are analyzed and extracted accordingly. Secondly, the Informer model is involved and then the CNN is used to extract the data features, and the GRU is utilized to capture the extracted data features. Finally, the proposed model is used to accomplish the accurate forecasting of short-term power load. It is shown that the RMSE values of the CNN-GRU-Informer model are only 0.0228 and 0.0258 in the A and B datasets, respectively. Therefore, the proposed model can well achieve the short-term power forecasting and provide a reference for other forecasting methods.