Issue |
E3S Web Conf.
Volume 522, 2024
2023 9th International Symposium on Vehicle Emission Supervision and Environment Protection (VESEP2023)
|
|
---|---|---|
Article Number | 01030 | |
Number of page(s) | 8 | |
DOI | https://doi.org/10.1051/e3sconf/202452201030 | |
Published online | 07 May 2024 |
Solar irradiance prediction model based on CNN-Bi-LSTM-Attention
School of Artificial Intelligence and Software, Liaoning Petrochemical University, Fushun, China
* Corresponding author: 1623052491@qq.com
The instability of solar irradiance can lead to power fluctuations in photovoltaic power generation systems, which can have a negative impact on the stability of the power system network. To address this problem, a new solar irradiance prediction model is proposed in this article, aiming to improve the stability of photovoltaic power generation systems and ensure the normal operation of the power grid network. Firstly, data pre-processing is performed to select input data, and then the data is fed into a convolutional neural network for fusion and feature extraction of multiple variables. The input of a bidirectional long short-term memory network model is set as the extracted results. Finally, the output of the network model is integrated into an attention mechanism to allocate weights and obtain the final prediction of solar irradiance. The predictive performance of the proposed model was validated using all-weather observation data from the publicly available online dataset in 2020. Compared with the control model, the proposed model showed significant improvements in average absolute error and other indicators.
Key words: Solar irradiation prediction / Eigenvalue processing / Convolutional neural network / Bidirectional long short-term memory network / Attention mechanism
© The Authors, published by EDP Sciences, 2024
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.