Reconstructing multi-orientation irradiance via PV panels: An isotropic Physical-MLP hybrid model

Yang Gao, Haoyu Ma, Jianan Zheng, Wenjun Liu, Fangcai Chen, Liulu Fan, Hanchun Wang, Wen Liu and Xinyu Zhang

Renewable Energy, 2025, vol. 241, issue C

Abstract: With the rapid development of solar energy, a large number of studies need solar irradiance data from different angles, while it is not feasible to collect arbitrary angles. To address this problem, we propose a hybrid model that uses two schemes with different characteristics to calculate solar irradiance from different angles and reconstruct the irradiance field. In this study, a small dataset located in Xiong'an New Area, China, containing various tilt angles and orientation angles is provided for model validation. We verify the performance of the two schemes, where the nRMSE of the Fixed-PV scheme is 6.75 % and the R2 value is 0.9171, while the nRMSE of the Tracking-PV scheme is 3.82 % and the R2 value is 0.981. In addition, we validated the performance of the model at different time resolutions (1s, 10s, 1min), different angles, and different weather, with the R2 value ranged from 0.8921 to 0.9171 for the Fixed-PV scheme, and from 0.9751 to 0.981 for the Tracking-PV scheme. The results show that the way of using the output of the physical model as the input parameter of machine learning improves the robustness of the hybrid model, overcoming the shortcomings of machine learning that can only predict solar irradiance based on a large amount of data. In addition, the two schemes are used to reconstruct the solar irradiance field to predict the single-day energy-receiving density from different angles. We used this model to predict the hourly power generation of two tilt-variable roof photovoltaic (PV) at Xiong'an station, which is the largest high-speed rail station in Asia. The results show that our predicted power generation follows the trend of the actual power generation, and the actual total power generation in different areas is about 85.66%–98.6 % of the predicted total power generation. The proposed method can be widely adopted to measure solar irradiance at other angles with PV panels as components. It is helpful to analyze and understand the optimal tilt angle of PV systems in the region and evaluate the power generation potential of PV systems on irregular building facades before implementing the project.

Keywords: Irradiance transposition; Solar irradiance calculation; Hybrid model; Machine learning; Photovoltaic potential estimation (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0960148125000084
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:renene:v:241:y:2025:i:c:s0960148125000084

DOI: 10.1016/j.renene.2025.122346

Access Statistics for this article

Renewable Energy is currently edited by Soteris A. Kalogirou and Paul Christodoulides

More articles in Renewable Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().