抄録
With the rapid evolution of power systems toward digitalization, networking, and intelligence, their structural complexity and operational dynamics have significantly increased. Traditional methods that rely on analytical physical models are increasingly constrained by difficulties in modeling, limited real-time capabilities, and poor adaptability in stability analysis and control. In recent years, the widespread deployment of phasor measurement units, intelligent terminals, and various sensors has provided high-dimensional and high-timeliness data support for power system operations, fostering a paradigm shift toward data-driven research methods. Artificial intelligence (AI) technologies—particularly machine learning, deep neural networks, and reinforcement learning—have been extensively explored for key tasks such as stability state identification, dynamic risk prediction, control strategy optimization, and adaptive fault response. This paper presents a comprehensive review of recent advances in the application of AI to power system stability, with a focus on data-driven dynamic stability prediction models, AI-integrated coordinated control strategies, and intelligent decision-making frameworks for disturbance detection and self-healing recovery. It further discusses core challenges in current research, including data quality and completeness, model generalization and robustness, algorithm interpretability, and feasibility of engineering deployment. Finally, the paper proposes key future research directions and potential breakthroughs. This paper aims to provide a systematic reference for advancing both theoretical research and practical implementation of AI technologies in power system stability assurance.
