A Review of Machine Learning Applications in Power System Emergency Control

Abstract

This study provides a comprehensive analysis of machine learning algorithms applied to emergency control in modern power systems. It explores the prospects for integrating machine learning (ML) into stability control systems and demonstrates that these algorithms can enable emergency responses based on the post-fault principle. The paper reviews methods for assessing stability and selecting control actions, establishing that the stability evaluation can be formulated as a binary classification problem, while the control action determination can be treated as a regression task. Additionally, it examines machine learning-based disturbance identification, with fault location serving as a trigger for automatic emergency response. Finally, it identifies promising research directions, including expanding the types of control actions and conducting comparative analyses of supervised and unsupervised machine learning applications.