Back to Publications

Spatiotemporal feature encoded deep learning method for rooftop PV potential assessment

XU Jian, Zhiling Guo, Qing Yu, DONG Kechuan, TAN Hongjun, Haoran Zhang, Jinyue Yan

DOI
Summary
A deep learning framework that encodes spatiotemporal data to predict shadow casting and rooftop PV potential (annual MAPE 2.85%, shading mIoU 89.23%), about 158 times faster than simulation.
Type
Journal Article
Publication
Applied Energy, 394, 126171
Spatiotemporal feature encoded deep learning method for rooftop PV potential assessment
Figure 1: Building heights and annual rooftop irradiation around a shaded building.
  • Solar energy
  • Deep learning
  • Urban energy

Cite this paper

@article{xu2025spatiotemporal,
  title   = {Spatiotemporal feature encoded deep learning method for rooftop PV potential assessment},
  author  = {Xu, Jian and Zhiling Guo and Qing Yu and Dong, Kechuan and Tan, Hongjun and Haoran Zhang and Jinyue Yan},
  journal = {Applied Energy},
  year    = {2025},
  doi     = {10.1016/j.apenergy.2025.126171}
}
Download .bib