Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes
Publicly accessible
License
This letter proposes a multi-task deep reinforcement learning (DRL) approach for distribution system voltage regulation considering topology changes via PV smart inverter control. The key idea is to encode the topology as an additional state for the DRL and leverage the multi-task learning scheme for joint learning of all task control policies. Unlike other DRL-based methods, our approach is robust to different topologies. Comparison results on the modified IEEE 123-node system demonstrate the enhanced robustness of the proposed method.
Citation Formats
TY - DATA
AB - This letter proposes a multi-task deep reinforcement learning (DRL) approach for distribution system voltage regulation considering topology changes via PV smart inverter control. The key idea is to encode the topology as an additional state for the DRL and leverage the multi-task learning scheme for joint learning of all task control policies. Unlike other DRL-based methods, our approach is robust to different topologies. Comparison results on the modified IEEE 123-node system demonstrate the enhanced robustness of the proposed method.
AU - Pei, Yansong
A2 - Zhao, Junbo
A3 - Yao, Yiyun
A4 - Ding, Fei
DB - C-MIX - Community Microgrid Information Exchange
DP - Open EI | National Laboratory of the Rockies
DO -
KW - Battery energy storage
KW - Solar
KW - Photovoltaics
KW - PV
KW - Diesel generators
KW - Other liquid-fuel generators
KW - Wind energy
KW - Power plant controls
KW - SCADA
KW - Policy and regulation
KW - Policy
KW - Regulation
KW - Case studies
KW - Performance
LA - English
DA - 2023/05/01
PY - 2023
PB - University of Conneticut
T1 - Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes
UR - https://cmix.openei.org/submissions/404
ER -
Pei, Yansong, et al. Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes. University of Conneticut, 1 May, 2023, C-MIX - Community Microgrid Information Exchange. https://cmix.openei.org/submissions/404.
Pei, Y., Zhao, J., Yao, Y., & Ding, F. (2023). Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes. [Data set]. C-MIX - Community Microgrid Information Exchange. University of Conneticut. https://cmix.openei.org/submissions/404
Pei, Yansong, Junbo Zhao, Yiyun Yao, and Fei Ding. Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes. University of Conneticut, May, 1, 2023. Distributed by C-MIX - Community Microgrid Information Exchange. https://cmix.openei.org/submissions/404
@misc{CMIX_Dataset_404,
title = {Multi-Task Reinforcement Learning for Distribution System Voltage Control With Topology Changes},
author = {Pei, Yansong and Zhao, Junbo and Yao, Yiyun and Ding, Fei},
abstractNote = {This letter proposes a multi-task deep reinforcement learning (DRL) approach for distribution system voltage regulation considering topology changes via PV smart inverter control. The key idea is to encode the topology as an additional state for the DRL and leverage the multi-task learning scheme for joint learning of all task control policies. Unlike other DRL-based methods, our approach is robust to different topologies. Comparison results on the modified IEEE 123-node system demonstrate the enhanced robustness of the proposed method.},
url = {https://cmix.openei.org/submissions/404},
year = {2023},
howpublished = {C-MIX - Community Microgrid Information Exchange, University of Conneticut, https://cmix.openei.org/submissions/404},
note = {Accessed: 2026-08-06}
}
Details
Data from May 1, 2023
Last updated Mar 30, 2026
Submitted Jun 2, 2026
Organization
University of Conneticut
Contact
Fei Ding
Authors
