ARTIFICIAL INTELLIGENCE IN ELECTRICAL SYSTEMS: A STRUCTURED NARRATIVE REVIEW OF FOUNDATIONS, APPLICATIONS, DIGITAL ARCHITECTURES AND CHALLENGES IN GRID MODERNIZATION

Abstract

The incorporation of Artificial Intelligence (AI) into electrical systems has become one of the central drivers of energy-sector digitalization. The growing operational complexity of power networks, the integration of intermittent renewable sources, the expansion of distributed energy resources, and the need for increasingly faster decisions have strengthened the relevance of machine learning, deep learning, expert systems, and reinforcement learning methods. This article presents a structured narrative review of the foundations, applications, and limitations of AI in electrical systems, with emphasis on load and renewable generation forecasting, predictive maintenance, fault diagnosis, operational optimization, and emerging digital architectures. Rather than providing a merely descriptive overview, the study organizes the literature into analytical axes, compares methodological families in terms of objectives, data requirements, strengths, and constraints, and discusses contemporary concepts such as cyber-physical energy systems, digital twins, distributed energy resource management, and edge intelligence. The results indicate that AI expands grid observability, automation capability, and anticipation of critical events, but large-scale deployment still depends on trustworthy data, interoperability, explainability, cybersecurity, and regulatory compliance. It is concluded that AI should not be understood only as an incremental automation tool, but as a structural component of the modernization of intelligent, resilient, and decentralized power grids.