Artificial Intelligence Applications in Power Systems: A Review of Methods, Use Cases, and Future Directions
Abstract
Modern electrical networks face growing operational demands as electricity consumption rises and renewable generation, distributed resources, and electric vehicles become more common. Artificial intelligence (AI) can learn useful relationships from operating records and simulated data, supporting tasks such as demand prediction, fault identification, state estimation, stability screening, and energy scheduling. This review organizes reported AI techniques and applications in both large interconnected grids and smaller power systems. It considers rule-based and fuzzy approaches alongside machine learning, deep learning, reinforcement learning, and newer foundation-model methods. The literature suggests that AI may accelerate selected analyses and provide flexible decision support, but dependable performance still relies on representative data, generalization, explainability, cyber protection, and real-world validation. A staged approach that combines AI with engineering models and expert oversight is outlined. This paper synthesizes existing studies and does not claim new experimental results..