Multi‐Agent Volt‐ VAR Control for Renewable Power Stations With Grid‐Forming Energy Storage: An Improved MASAC Approach
Engineering Reports · 2026-07 · 8 (7)
DOI: 10.1002/eng2.70835 ↗Kurz gefasst
Fachartikel in der Zeitschrift Engineering Reports, online erschienen am 09.07.2026. Verfasst von Kai Guan und 5 weiteren Autor:innen. Die Arbeit nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The voltage instability induced by the high‐penetration integration of renewable energy into power grids is a critical challenge. This article proposes an innovative coordinated control framework that synergizes grid‐forming energy storage (GFM‐ES) with multi‐agent reinforcement learning (MARL). The main innovations are threefold. First, an architectural innovation: a novel hierarchical “centralized‐distributed” architecture based on online updateable dynamic electrical zoning, which coordinates GFM‐ES (acting as a voltage‐forming master) and grid‐following resources (GFL) (acting as fast‐compensating slaves) across timescales. Second, an algorithmic innovation: an enhanced Multi‐Agent Soft Actor‐Critic (MASAC) algorithm featuring a multi‐timescale experience replay mechanism. This mechanism leverages ultra‐short‐term forecasting to adaptively partition operational periods and manage differentiated experience pools, enabling rapid, scenario‐aware learning under volatility. Third, a safety innovation: a standalone real‐time safety filter whose low computational overhead (< 10 ms) and minimal intervention rate (< 4%) are quantitatively validated to ensure secure deployment. It employs voltage prediction and convex‐optimization‐based correction to eliminate the risk of control‐induced voltage violations while preserving exploration, thus bridging the gap to real‐world deployment. Simulations verify superiority: the strategy maintains the lowest steady‐state voltage deviation (0.007–0.012 p.u.) across various weather conditions and achieves a 1.2‐s recovery without overshoot alongside a ˜98.2% voltage qualification rate under extreme events (e.g., 50% PV drop, 60% penetration). This work provides a control framework with demonstrated potential for safe, autonomous voltage control in modern renewable power stations.
Zitieren
APA
Guan, K., Xu, S., Li, Z., Li, Z., Xu, L., & Gao, P. (2026). Multi‐Agent Volt‐ VAR Control for Renewable Power Stations With Grid‐Forming Energy Storage: An Improved MASAC Approach. Engineering Reports, 8(7). https://doi.org/10.1002/eng2.70835
Vancouver
Guan K, Xu S, Li Z, Li Z, Xu L, Gao P. Multi‐Agent Volt‐ VAR Control for Renewable Power Stations With Grid‐Forming Energy Storage: An Improved MASAC Approach. Engineering Reports. 2026;8(7). doi:10.1002/eng2.70835
BibTeX
@article{Guan2026,
author = {Guan, Kai and Xu, Su and Li, Zhiling and Li, Zhibo and Xu, Lei and Gao, Pengfei},
title = {Multi‐Agent Volt‐ VAR Control for Renewable Power Stations With Grid‐Forming Energy Storage: An Improved MASAC Approach},
journal = {Engineering Reports},
year = {2026},
volume = {8},
number = {7},
doi = {10.1002/eng2.70835}
}
EUVIMED