← Back to Volume 11

Review on the Application of Reinforcement Learning in Distribution Network Voltage Control

Yanshen Zhao
pp.278-287 DOI: 10.62051/064r5j52 December 25, 2025
Download PDF
Abstract: Based on the core of Survey on the Application of Reinforcement Learning in Distribution Network Voltage Control, this study systematically organizes RL's application system in distribution network voltage control: it establishes MDP and POMDP modeling frameworks, clarifying the design logic of state, action, and reward functions; classifies and reviews core schemes of value function-based, policy gradient-based, and residual RL, quantitatively comparing algorithm performance via unified benchmarks and evaluation metrics; analyzes engineering implementation challenges (safety, scalability, interpretability) and proposes solutions (hierarchical control, Sim-to-Real transfer, human-machine collaboration); finally, it outlines short-term engineering paths and medium-term technical directions, providing references for intelligent voltage control of distribution networks in the new power system.
Keywords: Reinforcement Learning; Distribution Network; Voltage Control; DistFlow Model; Multi-Agent Collaboration; Residual Learning; Renewable Energy with High Penetration Rate.
APA Citation: Yanshen Zhao (2025). Review on the Application of Reinforcement Learning in Distribution Network Voltage Control. Transactions on Computer Science and Intelligent Systems Research, 11(1), 278-287. https://doi.org/10.62051/064r5j52

References

  1. Sun, Y. Z., & Wang, Z. F. (1998). Simulation model of OLTC and its impact on voltage and reactive power stability [J]. Automation of Electric Power Systems, 22 (5), 10–13.
  2. Han, X. M., Li, C. C., & Yang, X. (2025). Research on voltage control of distributed photovoltaic power generation systems connected to distribution networks [J]. New Energy Power Generation and Energy Storage, (5), 94–96.
  3. Luo, J. (2024). Research on deep reinforcement learning methods based on exploration and bias estimation [D]. Changchun: School of Computer Science and Technology, Jilin University. (Master’s Thesis in Engineering; Student ID: 2021534037; Unit Code: 10183; Classification: Public).
  4. Xiao, H., Wan, J., Xing, Y. B., et al. (2022). Power load forecasting strategy based on deep residual network [J]. Electric Engineering, (06), 1–4. https://doi.org/10.19768/j.cnki.dgjs.2022.06.039.
  5. Zheng, J. Y., Zhang, Z. H., Xuan, J. Q., et al. (2024). Intelligent planning method for distribution networks based on knowledge graph and graph convolutional neural network [J]. Computer Engineering, (Online First), 1–12.
  6. Alizadeh, B., Sheibani, M., Hashemi, S. M., & Marini, A. (2024). On the accuracy of linear DistFlow method: A comparison survey. In 2024 9th International Conference on Technology and Energy Management (ICTEM) (pp. 1–6). IEEE.
  7. Liu, W. (2025). Optimization of reactive power compensation technology in power capacitors [J]. Paper and Paper Machinery, 54 (2), 109–111.
  8. Zhang, F., Zhang, P. C., & Yang, H. (2025). Network loss/voltage optimization control of DC distribution networks based on proximal policy optimization algorithm [J]. Smart Grid, 43 (4), 75–84.
  9. Wang, X. H., Deng, J., & Yang, Z. X. (2020). Parameter optimization strategy for power system controllers [J]. Electric Machines and Control, 24 (9), 95–104.