Multi-Agent Reinforcement Learning for Cooperative Decision-Making in Power System Fault Diagnosis

Zhang Jifan

doi:10.23977/cpcs.2025.090106

Multi-Agent Reinforcement Learning for Cooperative Decision-Making in Power System Fault Diagnosis

Download as PDF

DOI: 10.23977/cpcs.2025.090106 | Downloads: 5 | Views: 418

Author(s)

Zhang Jifan ¹

Affiliation(s)

¹ Future Technology School, South China University of Technology, Guangzhou City, Guangdong Province, 510000, China

Corresponding Author

Zhang Jifan

ABSTRACT

Real-time and accurate fault diagnosis in power systems is crucial for ensuring grid safety and stability. Traditional centralized diagnostic methods struggle to cope with the dynamics and complexity of large-scale power networks, while single-agent reinforcement learning exhibits limitations in distributed collaborative decision-making. This paper proposes a cooperative fault diagnosis mechanism based on Multi-Agent Reinforcement Learning (MARL), enhancing fault localization accuracy and efficiency through distributed perception and optimized decision-making. First, a hierarchical collaborative architecture is constructed, comprising regional monitoring agents for local state perception and decision-coordinating agents for global optimization. Second, an attention mechanism-based information-sharing strategy and hybrid reward function are designed to address credit assignment under partial observability. Finally, a typical fault scenario library is established on the RTDS simulation platform for validation. Experimental results demonstrate significant advantages over conventional DDPG and independent Q-learning methods in diagnostic accuracy , response time , and generalization capability. This study provides a scalable distributed intelligent decision-making solution for power system fault diagnosis, offering practical significance for enhancing smart grid resilience.

KEYWORDS

Multi-Agent Reinforcement Learning; Power Systems; Fault Diagnosis; Cooperative Decision-Making; Distributed Perception; Attention Mechanism

CITE THIS PAPER

Zhang Jifan, Multi-Agent Reinforcement Learning for Cooperative Decision-Making in Power System Fault Diagnosis. Computing, Performance and Communication Systems (2025) Vol. 9: 36-40. DOI: http://dx.doi.org/10.23977/cpcs.2025.090106.

REFERENCES

[1] Liu, Z. Y. (2025). Research on the Application of Power System Automation Technology in Grid Operation Management. Science and Innovation, (10), 198-200+204.
[2] Chen, Q. G. (2025). Analysis of distribution line fault maintenance in power systems. In China Tendering Periodical Co., Ltd. (Ed.), Proceedings of the Forum on New Quality Productivity Driving Secondary Industry Development and Tendering Innovation (Vol. 2) (pp. 14-15).
[3] Chang, X. Y. & Wang, W. Q. (2025, April 23). Research on AI-based fault diagnosis and prediction in power systems. Market Information News, p. 014.

Subscription

E-Mail Alert

Downloads:	3202
Visits:	189671

Multi-Agent Reinforcement Learning for Cooperative Decision-Making in Power System Fault Diagnosis

Author(s)

Affiliation(s)

Corresponding Author

ABSTRACT

KEYWORDS

CITE THIS PAPER

REFERENCES

RESOURCES

JOIN US

PUBLICATION SERVICES

CONTACT US