Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding

Jizhe Liu; Yuchen Zhang; Ke Meng; Yan Xu; Zhao Yang Dong

doi:10.1109/ISGTAsia54193.2022.10003523

Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding

Jizhe Liu, Yuchen Zhang, Ke Meng, Yan Xu, Zhao Yang Dong

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

2 Citations (Scopus)

Abstract

With the increasing integration of renewable energy sources, maintaining secure and reliable power system operation becomes more challenging due to the escalated stochasticity and variations in the system. Emergency load shedding (ELS) serves as a fast and effective stability control scheme for power system after a risky disturbance occurs, which can suppress grid oscillation, recover system stability, and prevent cascading failure. Recently, data-driven techniques provide a new way to realize real-time ELS owing to their fast decision-making capability. This paper presents a risk-averse graph learning method for real-time ELS, where a graphSAGE model is proposed to fully capture the topology of power network and efficiently embed it into deep learning, and a risk-averse learning algorithm is used to avoid control failures induced by load under-cutting. The proposed method has been tested on New England 39-bus system and Nordic power system. The test results demonstrate the proposed ELS method can effectively reduce the overall control costs as compared to existing methods. © 2022 IEEE.

Original language	English
Title of host publication	Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022)
Publisher	IEEE
Pages	520-524
ISBN (Electronic)	9798350399660
ISBN (Print)	9798350399677
DOIs	https://doi.org/10.1109/ISGTAsia54193.2022.10003523
Publication status	Published - Nov 2022
Externally published	Yes
Event	11th International Conference on Innovative Smart Grid Technologies - Asia (ISGT-Asia 2022) - , Singapore Duration: 1 Nov 2022 → 5 Nov 2022

Publication series

Name	Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia
ISSN (Print)	2378-8534
ISSN (Electronic)	2378-8542

Conference

Conference	11th International Conference on Innovative Smart Grid Technologies - Asia (ISGT-Asia 2022)
Abbreviated title	IEEE ISGT-Asia 2022
Place	Singapore
Period	1/11/22 → 5/11/22

UN SDGs

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

SDG 7 Affordable and Clean Energy

Research Keywords

Deep learning
emergency load shedding
graph neural network
GraphSAGE
power system stability control

Access Output

10.1109/ISGTAsia54193.2022.10003523

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Liu, J., Zhang, Y., Meng, K., Xu, Y., & Dong, Z. Y. (2022). Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding. In Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022) (pp. 520-524). (Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia). IEEE. https://doi.org/10.1109/ISGTAsia54193.2022.10003523

@inproceedings{763c2861faf942709c873caf0d494e58,

title = "Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding",

abstract = "With the increasing integration of renewable energy sources, maintaining secure and reliable power system operation becomes more challenging due to the escalated stochasticity and variations in the system. Emergency load shedding (ELS) serves as a fast and effective stability control scheme for power system after a risky disturbance occurs, which can suppress grid oscillation, recover system stability, and prevent cascading failure. Recently, data-driven techniques provide a new way to realize real-time ELS owing to their fast decision-making capability. This paper presents a risk-averse graph learning method for real-time ELS, where a graphSAGE model is proposed to fully capture the topology of power network and efficiently embed it into deep learning, and a risk-averse learning algorithm is used to avoid control failures induced by load under-cutting. The proposed method has been tested on New England 39-bus system and Nordic power system. The test results demonstrate the proposed ELS method can effectively reduce the overall control costs as compared to existing methods. {\textcopyright} 2022 IEEE.",

keywords = "Deep learning, emergency load shedding, graph neural network, GraphSAGE, power system stability control",

author = "Jizhe Liu and Yuchen Zhang and Ke Meng and Yan Xu and Dong, {Zhao Yang}",

year = "2022",

month = nov,

doi = "10.1109/ISGTAsia54193.2022.10003523",

language = "English",

isbn = "9798350399677",

series = "Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia",

publisher = "IEEE",

pages = "520--524",

booktitle = "Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022)",

address = "United States",

note = "11th International Conference on Innovative Smart Grid Technologies - Asia (ISGT-Asia 2022), IEEE ISGT-Asia 2022 ; Conference date: 01-11-2022 Through 05-11-2022",

}

Liu, J, Zhang, Y, Meng, K, Xu, Y & Dong, ZY 2022, Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding. in Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022). Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia, IEEE, pp. 520-524, 11th International Conference on Innovative Smart Grid Technologies - Asia (ISGT-Asia 2022), Singapore, 1/11/22. https://doi.org/10.1109/ISGTAsia54193.2022.10003523

Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding. / Liu, Jizhe; Zhang, Yuchen; Meng, Ke et al.
Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022). IEEE, 2022. p. 520-524 (Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding

AU - Liu, Jizhe

AU - Zhang, Yuchen

AU - Meng, Ke

AU - Xu, Yan

AU - Dong, Zhao Yang

PY - 2022/11

Y1 - 2022/11

N2 - With the increasing integration of renewable energy sources, maintaining secure and reliable power system operation becomes more challenging due to the escalated stochasticity and variations in the system. Emergency load shedding (ELS) serves as a fast and effective stability control scheme for power system after a risky disturbance occurs, which can suppress grid oscillation, recover system stability, and prevent cascading failure. Recently, data-driven techniques provide a new way to realize real-time ELS owing to their fast decision-making capability. This paper presents a risk-averse graph learning method for real-time ELS, where a graphSAGE model is proposed to fully capture the topology of power network and efficiently embed it into deep learning, and a risk-averse learning algorithm is used to avoid control failures induced by load under-cutting. The proposed method has been tested on New England 39-bus system and Nordic power system. The test results demonstrate the proposed ELS method can effectively reduce the overall control costs as compared to existing methods. © 2022 IEEE.

AB - With the increasing integration of renewable energy sources, maintaining secure and reliable power system operation becomes more challenging due to the escalated stochasticity and variations in the system. Emergency load shedding (ELS) serves as a fast and effective stability control scheme for power system after a risky disturbance occurs, which can suppress grid oscillation, recover system stability, and prevent cascading failure. Recently, data-driven techniques provide a new way to realize real-time ELS owing to their fast decision-making capability. This paper presents a risk-averse graph learning method for real-time ELS, where a graphSAGE model is proposed to fully capture the topology of power network and efficiently embed it into deep learning, and a risk-averse learning algorithm is used to avoid control failures induced by load under-cutting. The proposed method has been tested on New England 39-bus system and Nordic power system. The test results demonstrate the proposed ELS method can effectively reduce the overall control costs as compared to existing methods. © 2022 IEEE.

KW - Deep learning

KW - emergency load shedding

KW - graph neural network

KW - GraphSAGE

KW - power system stability control

UR - http://www.scopus.com/inward/record.url?scp=85146833205&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85146833205&origin=recordpage

U2 - 10.1109/ISGTAsia54193.2022.10003523

DO - 10.1109/ISGTAsia54193.2022.10003523

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9798350399677

T3 - Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia

SP - 520

EP - 524

BT - Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022)

PB - IEEE

T2 - 11th International Conference on Innovative Smart Grid Technologies - Asia (ISGT-Asia 2022)

Y2 - 1 November 2022 through 5 November 2022

ER -

Liu J, Zhang Y, Meng K, Xu Y, Dong ZY. Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding. In Proceedings of the 11th International Conference on Innovative Smart Grid Technologies (Asia) (IEEE ISGT-Asia 2022). IEEE. 2022. p. 520-524. (Proceedings of the International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia). doi: 10.1109/ISGTAsia54193.2022.10003523

Risk-Averse Graph Learning for Real-time Power System Emergency Load Shedding

Abstract

Publication series

Conference

UN SDGs

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this