Review Article

Artificial Intelligence Applications in Smart Grids and Modern Power Systems: A Comprehensive Review

Abstract

The rapid digitalization of the power sector, coupled with increasing penetration of renewable energy sources and distributed energy resources, has fundamentally transformed the structure and operation of modern power systems. Traditional model-based and ruledriven approaches are increasingly challenged by the growing complexity, uncertainty, and scalability requirements of smart grids. In this context, artificial intelligence (AI) has emerged as a key enabling technology for enhancing forecasting accuracy, operational efficiency, system resilience, and cybersecurity. This paper presents a comprehensive review of AI applications in smart grids and modern power systems, covering methodological advances and practical implementations across multiple domains. A structured taxonomy of AI techniques-including machine learning, deep learning, reinforcement learning, and hybrid and explainable AI-is provided, along with a critical discussion of their suitability for power system applications. Key application areas such as load and renewable energy forecasting, power system operation and control, fault detection and predictive maintenance, demand response, distributed energy resource management, and grid resilience and security are systematically reviewed. Furthermore, the paper examines performance evaluation practices, benchmarking challenges, and deployment barriers related to data availability, scalability, interpretability, and regulatory constraints. Emerging research directions, including AI-driven digital twins, federated learning, foundation models, and AI-enabled net-zero grid operation, are also discussed. By synthesizing recent advances and identifying open research challenges, this review aims to provide valuable insights for researchers and practitioners seeking to develop reliable, scalable, and trustworthy AI solutions for next-generation power systems.