Open Access

ARTICLE

Safety Margin Guided Adaptive Coordinated Control for PV Storage Voltage Regulation in Active Distribution Networks

Baodong Li¹, Wei Lou^2,*, Chi Song¹, Qiang Gao¹, Rui Tang¹, Zhiming Dong^3,*
1 Chuzhou Power Supply Company, State Grid Anhui Electric Power Co., Ltd., Chuzhou, China
2 Electric Power Research Institute, State Grid Anhui Electric Power Co., Ltd., Hefei, China
3 School of Electrical Engineering and Automation, Hefei University of Technology, Hefei, China
* Corresponding Author: Wei Lou. Email: ; Zhiming Dong. Email:
(This article belongs to the Special Issue: Next-Generation Distribution System Planning, Operation, and Control)

Energy Engineering https://doi.org/10.32604/ee.2026.086652

Received 03 June 2026; Accepted 15 July 2026; Published online 31 August 2026

Abstract

High penetration of photovoltaic generation brings rapid voltage variation, uncertain reverse power flow, and increased network loss to active distribution networks. Voltage regulation becomes more challenging when PV output, load demand, and storage operation change simultaneously. To improve voltage security and operating economy under stochastic operating conditions, an adaptive coordinated control method named SMAC is proposed for PV storage voltage regulation. The control problem is formulated as a Markov decision process, where the operating state includes load demand, PV generation, storage state of charge, and voltage security information. The control action includes PV inverter reactive power, energy storage active power, and capacitor bank switching command. A safety margin guided reward function is constructed by integrating voltage deviation, voltage limit violation, active power loss, and soft low voltage risk. Different from violation driven control, the soft margin term provides early risk guidance before nodal voltages reach the lower security boundary. Within the actor critic framework, the actor learns online coordinated regulation actions, while the critic evaluates long term return related to voltage security and network loss. SMAC further introduces voltage security indicators, a safety-margin-guided reward function, and a hybrid coordinated action design for preventive voltage regulation under uncertain PV-load conditions. Case studies are conducted on a modified IEEE 33 bus distribution system with five PV units, one energy storage system, and one capacitor bank. Across five independent runs, SMAC achieves an average maximum voltage deviation of 0.0408 p.u. and increases the average minimum nodal voltage to 0.9592 p.u. No voltage violation is observed in any of the independent runs. The average network loss is reduced to 888.94 kWh. These statistical results demonstrate that SMAC provides stable preventive voltage regulation and favorable operating economy under different random initializations.

---

Keywords

distribution network; energy storage system; soft low-voltage security margin; deep reinforcement learning

---