EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Cross-Variable Time-Series Transformer Architecture Integrating Physical Features for Photovoltaic Energy Forecasting

Chen Xie, Mingju Chen, Yuyan Wang, Yangming Luo, Xueyang Duan, Zhihao Lin

Algorithms · 05.08.2026 · 19 (8) · 646

DOI: 10.3390/a19080646 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Algorithms, online erschienen am 05.08.2026. Verfasst von Chen Xie und 5 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Sichuan Provincial Natural Science Foundation, Opening Fund of the Intelligent Perception and Control Key Laboratory of Sichuan Province, Postgraduate Innovation Fund of Sichuan University of Science and Engineering.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate photovoltaic (PV) energy forecasting is vital for grid stability and the global low-carbon transition. However, existing data-driven and channel-independent PV energy forecasting models struggle to capture nonlinear meteorological couplings, heterogeneous physical scales across stations, and high-frequency non-stationary fluctuations. To address these limitations, this study proposes a Physics-Guided Cross-Variable Temporal Transformer architecture. Building upon a channel-independent foundation, we introduce a Cross-Variable Attention mechanism to explicitly reconstruct nonlinear photothermal couplings via dynamic attention weights. To resolve multi-station physical scale discrepancies, a Physical Feature-wise Linear Modulation network utilizes installed capacity as a static prior for adaptive cross-station scale alignment. During optimization, a Time Dynamics-Aware Perceiving Loss jointly penalizes absolute errors and first-order time differences, constraining the network’s tracking ability for transient ramping. Experiments demonstrate that the proposed architecture overcomes traditional channel-isolation limitations. The model achieves a 21.6% reduction in MSE compared to PatchTST, a 44.0% reduction compared to Autoformer, and a 2.7% improvement in R2 over Informer. This provides an accurate, generalizable, and physically interpretable solution for collaborative multi-station distributed PV energy forecasting.

Förderung

  • Sichuan Provincial Natural Science Foundation
  • Opening Fund of the Intelligent Perception and Control Key Laboratory of Sichuan Province
  • Postgraduate Innovation Fund of Sichuan University of Science and Engineering

Zitieren

APA

Xie, C., Chen, M., Wang, Y., Luo, Y., Duan, X., & Lin, Z. (2026). A Cross-Variable Time-Series Transformer Architecture Integrating Physical Features for Photovoltaic Energy Forecasting. Algorithms, 19(8), 646. https://doi.org/10.3390/a19080646

Vancouver

Xie C, Chen M, Wang Y, Luo Y, Duan X, Lin Z. A Cross-Variable Time-Series Transformer Architecture Integrating Physical Features for Photovoltaic Energy Forecasting. Algorithms. 2026;19(8):646. doi:10.3390/a19080646

BibTeX
@article{Xie2026,
  author = {Xie, Chen and Chen, Mingju and Wang, Yuyan and Luo, Yangming and Duan, Xueyang and Lin, Zhihao},
  title = {A Cross-Variable Time-Series Transformer Architecture Integrating Physical Features for Photovoltaic Energy Forecasting},
  journal = {Algorithms},
  year = {2026},
  volume = {19},
  number = {8},
  pages = {646},
  doi = {10.3390/a19080646}
}
Kostenlos starten →