Academic Journal of Engineering and Technology Science, 2025, 8(4); doi: 10.25236/AJETS.2025.080411.
Liyu Wu, Yanxin Liu, Xinyu Gao
School of Communication and Artificial Intelligence, School of Integrate Circuits, Nanjing Institute of Technology, Nanjing, China
This paper proposes an improved Transformer-based model MVTformer for multivariate time series forecasting tasks. The model introduces a sparse self-attention mechanism and a time series feature extraction module to improve the modeling capabilities of cross-variable dependencies and long-term dynamic features. MVTformer implemented on the PyTorch platform has good computational efficiency and scalability, and is suitable for complex industrial time series data scenarios. The experiment was carried out on a real wind power dataset. Compared with mainstream models such as LSTM, GRU, TCN and standard Transformer, MVTformer performed best in multiple evaluation indicators, fully verifying its accuracy and robustness in sequence prediction.
Machine Learning; Deep Learning; Time Series Modeling; Transformer Architecture; Self-Attention Mechanism; Wind Power Forecasting; Sequence Learning
Liyu Wu, Yanxin Liu, Xinyu Gao. Multi-Variable Time Series Forecasting for Wind Power Using Improved Transformer Architecture. Academic Journal of Engineering and Technology Science (2025), Vol. 8, Issue 4: 79-86. https://doi.org/10.25236/AJETS.2025.080411.
[1] D. Yang, B. Gao, S. Wang, and H. Xiang, “Robustness test for fouling state identification of homogeneous pressure electrodes based on confidence ellipsoids,” IEICE Electron. Express, p. 22. 20240283, 2025, doi: 10.1587/elex.22.20240283.
[2] E. Arslan Tuncar, Ş. Sağlam, and B. Oral, “A review of short-term wind power generation forecasting methods in recent technological trends,” Energy Reports, vol. 12, pp. 197–209, Dec. 2024, doi: 10.1016/j. egyr. 2024.06.006.
[3] N. Li, J. Dong, L. Liu, H. Li, and J. Yan, “A novel EMD and causal convolutional network integrated with transformer for ultra short-term wind power forecasting,” Int. J. Electr. Power Energy Syst., vol. 154, p. 109470, Dec. 2023, doi: 10.1016/j.ijepes.2023.109470.
[4] S. Wang, J. Shi, W. Yang, and Q. Yin, “High and low frequency wind power prediction based on transformer and BiGRU-attention,” Energy, vol. 288, p. 129753, Feb. 2024, doi: 10.1016/j.energy. 2023. 129753.
[5] J. Cheng, X. Luo, and Z. Jin, “Integrating domain knowledge into transformer for short-term wind power forecasting,” Energy, vol. 312, p. 133511, Dec. 2024, doi: 10.1016/j.energy.2024.133511.
[6] T. B. M. J. Ouarda and F. Houndekindo, “LSTM and transformer-based framework for bias correction of ERA5 hourly wind speeds,” 2025, SSRN. doi: 10.2139/ssrn.5125439.
[7] H. Zhou, “Informer: beyond efficient transformer for long sequence time-series forecasting,” Proc. AAAI Conf. Artif. Intell., vol. 35, no. 12, Art. no. 12, May 2021, doi: 10.1609/aaai.v35i12.17325.
[8] H. Eskandari, M. Imani, and M. P. Moghaddam, “A deep residual network integrating entropy-based wavelet packet ensemble model for short-term electrical load forecasting,” Energy, vol. 314, p. 134168, Jan. 2025, doi: 10.1016/j.energy.2024.134168.
[9] A. Ijadi Maghsoodi, A. E. Torkayesh, L. C. Wood, E. Herrera-Viedma, and K. Govindan, “A machine learning driven multiple criteria decision analysis using LS-SVM feature elimination: sustainability performance assessment with incomplete data,” Eng. Appl. Artif. Intell., vol. 119, p. 105785, Mar. 2023, doi: 10.1016/j.engappai.2022.105785.
[10] S. Abolhosseini, M. Khorashadizadeh, M. Chahkandi, and M. Golalizadeh, “A modified ID3 decision tree algorithm based on cumulative residual entropy,” Expert Syst. Appl., vol. 255, p. 124821, Dec. 2024, doi: 10.1016/j.eswa.2024.124821.
[11] E. G. S. Nascimento, T. A. C. De Melo, and D. M. Moreira, “A transformer-based deep neural network with wavelet transform for forecasting wind speed and wind energy,” Energy, vol. 278, p. 127678, Sep. 2023, doi: 10.1016/j.energy.2023.127678.
[12] A. Redekar, H. S. Dhiman, D. Deb, and S. M. Muyeen, “On reliability enhancement of solar PV arrays using hybrid SVR for soiling forecasting based on WT and EMD decomposition methods,” Ain Shams Eng. J., vol. 15, no. 6, p. 102716, Jun. 2024, doi: 10.1016/j.asej.2024.102716.
[13] M. Barman and N. B. Dev Choudhury, “Season specific approach for short-term load forecasting based on hybrid FA-SVM and similarity concept,” Energy, vol. 174, pp. 886–896, May 2019, doi: 10.1016/j.energy.2019.03.010.
[14] A. Dupré, P. Drobinski, B. Alonzo, J. Badosa, C. Briard, and R. Plougonven, “Sub-hourly forecasting of wind speed and wind energy,” Renewable Energy, vol. 145, pp. 2373–2379, Jan. 2020, doi: 10.1016/j.renene.2019.07.161.
[15] J. G. Martin, J. R. D. Frejo, J. M. Maestre, and E. F. Camacho, “Spatio-temporal kriging for spatial irradiance estimation with short-term forecasting in a thermosolar power plant,” Heliyon, vol. 10, no. 20, p. e39247, Oct. 2024, doi: 10.1016/j.heliyon.2024.e39247.