Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification

Huayi Wu; Zhao Xu; Yibo Ding; Wenzhuo Shi; Aoxiang Zhang; Yugui Liu

doi:10.1109/CICED63421.2024.10754175

Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification

Huayi Wu
, Zhao Xu
, Yibo Ding
, Wenzhuo Shi
, Aoxiang Zhang
, Yugui Liu

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

1 Citation (Scopus)

Abstract

Distribution system topology identification is the basic function for system operation but is hindered by data privacy concerns and uncertainties arising from high penetration of renewable energy sources (RES). To tackle these issues, a graph convolutional network-based federated learning method (GraphFed) is proposed for topology identification while ensuring data privacy. Specifically, to remove the data leakage risk, the federated learning (FL) algorihm is proposed for topology identification in a decentralized learning framework. Besides, the proposed graph convolutional network (GCN) leverages a novel node embedding-based graph shift operator to automatically represent the graphical features of the power system data to achieve enhanced topology identification accuracy. Comparative experiments are conducted on the IEEE 33-node distribution system, demonstrating the effectiveness of the proposed GraphFed method.

Original language	English
Title of host publication	Proceedings - 11th China International Conference on Electricity Distribution
Subtitle of host publication	More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024
Publisher	IEEE Computer Society
Pages	724-728
Number of pages	5
ISBN (Electronic)	9798350368345
DOIs	https://doi.org/10.1109/CICED63421.2024.10754175
Publication status	Published - Nov 2024
Event	11th China International Conference on Electricity Distribution, CICED 2024 - Hangzhou, China Duration: 12 Sept 2024 → 13 Sept 2024

Publication series

Name	China International Conference on Electricity Distribution, CICED
ISSN (Print)	2161-7481
ISSN (Electronic)	2161-749X

Conference

Conference	11th China International Conference on Electricity Distribution, CICED 2024
Country/Territory	China
City	Hangzhou
Period	12/09/24 → 13/09/24

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

Data privacy
distribution system topology identification
federated learning
graph convolutional network
node embedding

ASJC Scopus subject areas

Energy Engineering and Power Technology
Control and Systems Engineering
Electrical and Electronic Engineering

More information

10.1109/CICED63421.2024.10754175

Cite this

Wu, H., Xu, Z., Ding, Y., Shi, W., Zhang, A., & Liu, Y. (2024). Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification. In Proceedings - 11th China International Conference on Electricity Distribution: More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024 (pp. 724-728). (China International Conference on Electricity Distribution, CICED). IEEE Computer Society. https://doi.org/10.1109/CICED63421.2024.10754175

Wu, Huayi ; Xu, Zhao ; Ding, Yibo et al. / Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification. Proceedings - 11th China International Conference on Electricity Distribution: More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024. IEEE Computer Society, 2024. pp. 724-728 (China International Conference on Electricity Distribution, CICED).

@inproceedings{652e531946a94da28259a9d0ab9bacbe,

title = "Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification",

abstract = "Distribution system topology identification is the basic function for system operation but is hindered by data privacy concerns and uncertainties arising from high penetration of renewable energy sources (RES). To tackle these issues, a graph convolutional network-based federated learning method (GraphFed) is proposed for topology identification while ensuring data privacy. Specifically, to remove the data leakage risk, the federated learning (FL) algorihm is proposed for topology identification in a decentralized learning framework. Besides, the proposed graph convolutional network (GCN) leverages a novel node embedding-based graph shift operator to automatically represent the graphical features of the power system data to achieve enhanced topology identification accuracy. Comparative experiments are conducted on the IEEE 33-node distribution system, demonstrating the effectiveness of the proposed GraphFed method.",

keywords = "Data privacy, distribution system topology identification, federated learning, graph convolutional network, node embedding",

author = "Huayi Wu and Zhao Xu and Yibo Ding and Wenzhuo Shi and Aoxiang Zhang and Yugui Liu",

note = "Publisher Copyright: {\textcopyright} 2024 IEEE.; 11th China International Conference on Electricity Distribution, CICED 2024 ; Conference date: 12-09-2024 Through 13-09-2024",

year = "2024",

month = nov,

doi = "10.1109/CICED63421.2024.10754175",

language = "English",

series = "China International Conference on Electricity Distribution, CICED",

publisher = "IEEE Computer Society",

pages = "724--728",

booktitle = "Proceedings - 11th China International Conference on Electricity Distribution",

address = "United States",

}

Wu, H, Xu, Z, Ding, Y, Shi, W, Zhang, A & Liu, Y 2024, Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification. in Proceedings - 11th China International Conference on Electricity Distribution: More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024. China International Conference on Electricity Distribution, CICED, IEEE Computer Society, pp. 724-728, 11th China International Conference on Electricity Distribution, CICED 2024, Hangzhou, China, 12/09/24. https://doi.org/10.1109/CICED63421.2024.10754175

Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification. / Wu, Huayi; Xu, Zhao; Ding, Yibo et al.
Proceedings - 11th China International Conference on Electricity Distribution: More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024. IEEE Computer Society, 2024. p. 724-728 (China International Conference on Electricity Distribution, CICED).

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

TY - GEN

T1 - Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification

AU - Wu, Huayi

AU - Xu, Zhao

AU - Ding, Yibo

AU - Shi, Wenzhuo

AU - Zhang, Aoxiang

AU - Liu, Yugui

PY - 2024/11

Y1 - 2024/11

N2 - Distribution system topology identification is the basic function for system operation but is hindered by data privacy concerns and uncertainties arising from high penetration of renewable energy sources (RES). To tackle these issues, a graph convolutional network-based federated learning method (GraphFed) is proposed for topology identification while ensuring data privacy. Specifically, to remove the data leakage risk, the federated learning (FL) algorihm is proposed for topology identification in a decentralized learning framework. Besides, the proposed graph convolutional network (GCN) leverages a novel node embedding-based graph shift operator to automatically represent the graphical features of the power system data to achieve enhanced topology identification accuracy. Comparative experiments are conducted on the IEEE 33-node distribution system, demonstrating the effectiveness of the proposed GraphFed method.

AB - Distribution system topology identification is the basic function for system operation but is hindered by data privacy concerns and uncertainties arising from high penetration of renewable energy sources (RES). To tackle these issues, a graph convolutional network-based federated learning method (GraphFed) is proposed for topology identification while ensuring data privacy. Specifically, to remove the data leakage risk, the federated learning (FL) algorihm is proposed for topology identification in a decentralized learning framework. Besides, the proposed graph convolutional network (GCN) leverages a novel node embedding-based graph shift operator to automatically represent the graphical features of the power system data to achieve enhanced topology identification accuracy. Comparative experiments are conducted on the IEEE 33-node distribution system, demonstrating the effectiveness of the proposed GraphFed method.

KW - Data privacy

KW - distribution system topology identification

KW - federated learning

KW - graph convolutional network

KW - node embedding

UR - https://www.scopus.com/pages/publications/85212069383

U2 - 10.1109/CICED63421.2024.10754175

DO - 10.1109/CICED63421.2024.10754175

M3 - Conference article published in proceeding or book

AN - SCOPUS:85212069383

T3 - China International Conference on Electricity Distribution, CICED

SP - 724

EP - 728

BT - Proceedings - 11th China International Conference on Electricity Distribution

PB - IEEE Computer Society

T2 - 11th China International Conference on Electricity Distribution, CICED 2024

Y2 - 12 September 2024 through 13 September 2024

ER -

Wu H, Xu Z, Ding Y, Shi W, Zhang A, Liu Y. Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification. In Proceedings - 11th China International Conference on Electricity Distribution: More Reliable, More Flexible, and More Intelligent Distribution System, CICED 2024. IEEE Computer Society. 2024. p. 724-728. (China International Conference on Electricity Distribution, CICED). doi: 10.1109/CICED63421.2024.10754175

Graph Convolutional Network-Based Federated Learning Method for Distribution System Topology Identification

Abstract

Publication series

Conference

UN SDGs

Keywords

ASJC Scopus subject areas

More information

Other files and links

Fingerprint

Cite this