Two-Timescale Coordinated Voltage Regulation for High Renewable-Penetrated Active Distribution Networks Considering Hybrid Devices

The integration of large-scale distributed generators into active distribution networks (ADNs) will aggravate voltage fluctuations, which can affect the secure operation of power grids seriously. In this article, we investigate a cooperated voltage regulation problem of ADNs. Specifically, we first...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on industrial informatics 2024-03, Vol.20 (3), p.3456-3467
Main Authors: Zhang, Tingjun, Yu, Liang, Yue, Dong, Dou, Chunxia, Xie, Xiangpeng, Hancke, Gerhard P.
Format: Article
Language:English
Subjects:
Active distribution networks (ADNs)
Algorithms
bilevel Markov games
Coordination
Distributed generation
Electric potential
Games
hierarchical multi-agent attention-based deep reinforcement learning (HMAADRL)
hybrid devices
Machine learning
Markov processes
Multiagent systems
Optimization
Reactive power
Reagents
Regulation
Static VAr compensators
Time
two-timescale voltage regulation
Voltage
Voltage control
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The integration of large-scale distributed generators into active distribution networks (ADNs) will aggravate voltage fluctuations, which can affect the secure operation of power grids seriously. In this article, we investigate a cooperated voltage regulation problem of ADNs. Specifically, we first formulate a two-timescale voltage regulation problem considering the coordination of various hybrid devices while reducing the power loss of the whole ADNs. Given that the aforementioned problem is challenging to solve directly, we reformulate it as bilevel Markov games. Then, we propose a hierarchical multi-agent attention-based deep reinforcement learning algorithm to solve them. To be specific, the upper level Markov game is solved by a discrete multi-actor-attention-critic (MAAC) algorithm, and the lower level Markov game is solved by a continuous MAAC algorithm. In addition, the two-timescale coordination between upper level and lower level agents is implemented through the information exchange of rewards during the training process. Simulation results show that the proposed algorithm has good effectiveness, robustness, and scalability in voltage regulation.
ISSN:1551-3203
1941-0050
DOI:10.1109/TII.2023.3308348