A two-stage learning approach for wind power prediction using dynamic graph convolution network hybrid model
This paper proposes an effective two-stage learning method to improve the accuracy of wind power prediction. The complex time dependency relationship between meteorological data variables is modelled using a dynamic graph convolutional network, and the wind power prediction is completed using the XGBoost model. Experimental results show that the two-stage learning method can effectively extract key time series features and has good prediction accuracy.
Authors
- Shijie Guan (ORCID: https://orcid.org/0000-0003-4217-1871)
- Daorina Bao
- Xiaojun Liu (ORCID: https://orcid.org/0009-0006-7845-168X)
- Hailong Li (ORCID: https://orcid.org/0000-0002-6279-4446)
- Haotian Chen (ORCID: https://orcid.org/0000-0003-1788-6605)
- Yongsheng Wang (ORCID: https://orcid.org/0000-0002-4811-1346)
- Xuehui Wang
Institutions
- Dalian University of Technology (CN)
- Inner Mongolia Autonomous Region Meteorological Bureau (CN)
- China Agricultural University (CN)
- Inner Mongolia University of Technology (CN)
Publication Details
- Published
- 2024-11-29
- DOI
- https://doi.org/10.1109/ei264398.2024.10991430
- Citations
- 1
- Primary Topic
- Energy Load and Power Forecasting
- Type
- article
- Field-Weighted Citation Impact
- 0.21
Funders
- National Natural Science Foundation of China