EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Thermal-Aware Safe Reinforcement Learning for Cold-Storage-Assisted Voltage Control in Rural PV Microgrids

Yunfei Zhang, Kefan Bao, Hui Yan, Xuehong Yuan, Wennan Zhuang, Difei Tang, Xiangyu Lu, Guang Chen

Eng · 15.11.2026 · 7 (11) · 647

DOI: 10.3390/eng7110647 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Eng, online erschienen am 15.11.2026. Verfasst von Yunfei Zhang und 7 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: State Grid Jiangsu Electric Power Company.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Surplus photovoltaic (PV) generation can cause sustained overvoltage in rural feeders that inverter reactive-power control alone may not resolve. Agricultural cold storage can absorb surplus generation through refrigeration load shifting, but its grid support must preserve product temperatures and allow timely thermal recovery. This study proposes a thermal-aware safe reinforcement learning framework that coordinates cold-storage demand with inverter voltage regulation. A three-state air–product–envelope model and a shadow baseline quantify stored cooling and recoverable refrigeration flexibility. A Soft Actor–Critic policy uses electrical and thermal states with forecasts to schedule refrigeration, while predictive action mapping confines its commands to a recovery-feasible interval. A two-time-scale architecture coordinates this slower demand response with fast inverter voltage–reactive-power (Volt–VAR) control. In 90-day simulations on modified Institute of Electrical and Electronics Engineers (IEEE) 33- and 69-bus feeders, the framework reduces mean integrated overvoltage on common event days by 5.84% and 10.02%, respectively, relative to Volt–VAR, with no PV curtailment and all recoveries completed on time. These findings support the use of existing cold-storage thermal capacity to supplement inverter regulation and accommodate PV generation in rural grids.

Förderung

  • State Grid Jiangsu Electric Power Company

Zitieren

APA

Zhang, Y., Bao, K., Yan, H., Yuan, X., Zhuang, W., Tang, D., Lu, X., & Chen, G. (2026). Thermal-Aware Safe Reinforcement Learning for Cold-Storage-Assisted Voltage Control in Rural PV Microgrids. Eng, 7(11), 647. https://doi.org/10.3390/eng7110647

Vancouver

Zhang Y, Bao K, Yan H, Yuan X, Zhuang W, Tang D, et al. Thermal-Aware Safe Reinforcement Learning for Cold-Storage-Assisted Voltage Control in Rural PV Microgrids. Eng. 2026;7(11):647. doi:10.3390/eng7110647

BibTeX
@article{Zhang2026,
  author = {Zhang, Yunfei and Bao, Kefan and Yan, Hui and Yuan, Xuehong and Zhuang, Wennan and Tang, Difei and Lu, Xiangyu and Chen, Guang},
  title = {Thermal-Aware Safe Reinforcement Learning for Cold-Storage-Assisted Voltage Control in Rural PV Microgrids},
  journal = {Eng},
  year = {2026},
  volume = {7},
  number = {11},
  pages = {647},
  doi = {10.3390/eng7110647}
}
Kostenlos starten →