EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Day-Ahead Scheduling of Distributed Energy Resources in Net-Zero Energy Houses With Multi-Day Operation Using Soft Actor-Critic

Chao-Hsien Hsiao, Wei-Chen Lin, Wei-Tzer Huang

IEEE Access · 2026 · 14 · 69841-69860

DOI: 10.1109/access.2026.3690577 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Chao-Hsien Hsiao und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science and Technology Council.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a reinforcement learning–based day-ahead scheduling framework for distributed energy resources in a net-zero energy house equipped with photovoltaic panels, wind turbines, a battery energy storage systems, and optional electric vehicle flexibility. The net-zero energy house was connected to a low-voltage grid (three-phase four-wire 380/220 V and single-phase three-wire 220/110 V). A soft actor–critic controller generates continuous 24-hour schedules at 15-minute intervals while enforcing converter power limits and state-of-charge constraints. A cross-day mechanism carries the terminal state of charge of each day into the initial state of the next day, ensuring physically consistent daily rollouts without artificial State-of-Charge (SoC) resets. The reward formulation is designed to reduce grid imports and enhance the temporal matching between household load and distributed energy resource generation through netload shaping and state-of-charge-safe operation, with feasibility ensured via action projection under energy headroom constraints. The soft actor–critic policy is trained using the measured 2023 operation profiles and evaluated on the measured 2024 winter (January) and summer (July) data under two asset configurations: (i) a 30-kWh battery energy storage system without an electric vehicle and (ii) a 16-kWh battery energy storage system with electric vehicle support. Compared with the original measured netload operation, the proposed scheduler increases the average self-sufficiency rate to approximately 0.72 and the self-consumption rate to approximately 0.54, while reducing the total grid-import energy by more than 50% across both seasons and configurations, demonstrating stable month-long daily rollout performance with cross-day SoC carry-over for practical net-zero energy house deployment.

Förderung

  • National Science and Technology Council

Zitieren

APA

Hsiao, C., Lin, W., & Huang, W. (2026). Day-Ahead Scheduling of Distributed Energy Resources in Net-Zero Energy Houses With Multi-Day Operation Using Soft Actor-Critic. IEEE Access, 14, 69841-69860. https://doi.org/10.1109/access.2026.3690577

Vancouver

Hsiao C, Lin W, Huang W. Day-Ahead Scheduling of Distributed Energy Resources in Net-Zero Energy Houses With Multi-Day Operation Using Soft Actor-Critic. IEEE Access. 2026;14:69841-69860. doi:10.1109/access.2026.3690577

BibTeX
@article{Hsiao2026,
  author = {Hsiao, Chao-Hsien and Lin, Wei-Chen and Huang, Wei-Tzer},
  title = {Day-Ahead Scheduling of Distributed Energy Resources in Net-Zero Energy Houses With Multi-Day Operation Using Soft Actor-Critic},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {69841--69860},
  doi = {10.1109/access.2026.3690577}
}
Kostenlos starten →