Research on Photovoltaic Power Generation Energy Forecasting Model Driven by Deep Learning
DOI:
https://doi.org/10.54097/6z0p1a57Keywords:
Deep Learning, Photovoltaic Power Generation Forecasting, LSTM, CNN-LSTM, Model ComparisonAbstract
To improve the accuracy of photovoltaic power generation forecasting, this paper constructs two deep learning forecasting models: LSTM and CNN-LSTM. Historical power generation and meteorological data from photovoltaic power plants are collected and preprocessed for model training and optimization. Comparative experiments are conducted with models such as ARIMA, CNN, and GRU. Results show that the CNN-LSTM hybrid model achieves the best performance, with a MAE of 2.35 kW·h, an RMSE of 3.26 kW·h, a MAPE as low as 3.65%, and an R² of 0.958, significantly outperforming both ARIMA (MAE 6.85 kW·h, MAPE 10.56%) and a single LSTM (MAE 3.56 kW·h, MAPE 5.92%). Simulation curves demonstrate that the model accurately captures power generation fluctuations, particularly during peak and valley periods. Research has shown that a hybrid model that integrates the spatial characteristics of meteorological data with the temporal characteristics of power generation can effectively improve forecasting accuracy and provide a reliable basis for power system scheduling.
Downloads
References
[1] Chen, M., Zhu, Y., Sun, Y., Xie, Z. & Wu, P. Comprehensive energy system forecasting and control considering high-penetration photovoltaic consumption and deep reinforcement learning. Transactions of the Chinese Society of Electrical Engineering, Vol. 39(2024) No. 19, p. 6056071.
[2] Luo, X., Hu, Z., Ma, Z. & Lü, Z. Photovoltaic cluster power generation forecasting based on hybrid deep learning. Journal of Southwest University (Natural Science Edition), Vol. 47(2025) No. 3, p. 200-210.
[3] Tan, Y., Huang, Y., Liu, J. & Yang, R. Long-term and short-term time series forecasting of multivariate loads in power systems based on MMoE-CNN-Informer model. Transactions of the Chinese Society of Electrical Engineering, Vol. 20(2025) No. 2, p. 253-263.
[4] Han, J., Yan, L., Zhang, F., Wang, Y. & Mao, H. Research on the technical path of high-quality development of photovoltaic industry driven by digital-physical integration. Laser Journal, Vol. 45(2024) No. 11, p. 193-196.
[5] Wei, W., Lü, Y., Qi, X., Liu, J. & Fang, F. State prediction method of wind turbine in power station based on CNN-LSTM-AM dynamic integration model. Chinese Journal of Scientific Instrument, Vol. 44(2023) No. 4, p. 19-27.
[6] Zhang, S., Wang, J., Wang, C., Zhao, C. & Zhang, Y. Energy prediction of edge photovoltaic power generation based on hierarchical fuzzy neural network. Modern Electric Power, Vol. 41(2024) No. 3, p. 490-499.
[7] Fan, L., Zhang, Q., Li, G. & Wu, J. Short-term photovoltaic power generation prediction based on long short-term memory network digital twin. Modern Electric Power, Vol. 40(2023) No. 6, p. 899-905.
[8] Zhu, J., Dong, H., Li, S., Chen, Z. & Luo, T. A review of data-driven integrated energy system load forecasting. Proceedings of the CSEE, Vol. 41(2021) No. 23, p. 7905-7923.
[9] Zhao, Y., Wang, H., Kang, L. & Zhang, Z. Short-term power load forecasting based on temporal convolutional network. Transactions of the Chinese Society of Electrotechnical Engineering, Vol. 37(2022) No. 5, p. 1242-1251.
[10] Feng, P., Jiang, G., Xu, J., Ye, J. & Li, S. Short-term photovoltaic power forecasting based on CEEMD-BiLSTM-RFR. Science, Technology and Engineering, Vol. 24(2024) No. 5, p. 1955-1962.
[11] Zheng, X., Yang, X. & Lai, W. Classification and research progress of renewable energy power generation forecasting methods. Science, Technology and Engineering, Vol. 24(2024) No. 34, p. 14503-14521.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
