EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Physics-Informed Neural Network of Half-Inverse Gradient Method for Solving the Power Flow

Zhencheng Liang, Zonglong Weng, Biyun Chen, Bin Li, Peijie Li

Sustainability · 29.04.2026 · 18 (9) · 4386

DOI: 10.3390/su18094386 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sustainability, online erschienen am 29.04.2026. Verfasst von Zhencheng Liang und 4 weiteren Autor:innen. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China Southern Power Grid Co., Ltd..

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Power flow (PF) analysis is fundamental for power system operation and planning, yet traditional methods like Newton–Raphson face problems in convergence and computational efficiency. While deep learning (DL) offers promising solutions, its “black-box” nature and unstable training dynamics hinder practical adoption. This paper proposes a physics-informed neural network (PINN) framework integrated with a novel half-inverse gradient (HIG) mechanism to address these limitations. First, a systematic study of gradient scaling in PF optimisation found that the lack of enough inverse matrix compensation was the main cause of training instability. Second, we design a residual-driven HIG method that compensates gradient matrices via inverse operations, enabling accelerated convergence while maintaining numerical stability. Third, we develop parameterized voltage variables with differentiable activation functions to enforce hard operational constraints. The HIG optimizer leverages automatic differentiation and truncated singular value decomposition to balance diagonal/non-diagonal gradient information, achieving 99% accuracy in case4gs and case30 studies. Experiments on case118 demonstrate the framework’s scalability, with 65% accuracy compared to about 38% for baseline physics-informed approaches.

Förderung

  • China Southern Power Grid Co., Ltd.

Zitieren

APA

Liang, Z., Weng, Z., Chen, B., Li, B., & Li, P. (2026). Physics-Informed Neural Network of Half-Inverse Gradient Method for Solving the Power Flow. Sustainability, 18(9), 4386. https://doi.org/10.3390/su18094386

Vancouver

Liang Z, Weng Z, Chen B, Li B, Li P. Physics-Informed Neural Network of Half-Inverse Gradient Method for Solving the Power Flow. Sustainability. 2026;18(9):4386. doi:10.3390/su18094386

BibTeX
@article{Liang2026,
  author = {Liang, Zhencheng and Weng, Zonglong and Chen, Biyun and Li, Bin and Li, Peijie},
  title = {Physics-Informed Neural Network of Half-Inverse Gradient Method for Solving the Power Flow},
  journal = {Sustainability},
  year = {2026},
  volume = {18},
  number = {9},
  pages = {4386},
  doi = {10.3390/su18094386}
}
Kostenlos starten →