From Algorithms to Applications: Data-Centric Optimization for Distributed Power System Energy Management

Abstract

With the rapid advancement of computer-driven optimization and artificial intelligence, energy management in distributed power systems (DPS) has emerged as a critical research focus in the power and energy sector. This review systematically summarizes recent progress in optimization algorithms for distributed power systems, highlighting their theoretical foundations, methodological diversity, and practical applications. First, traditional mathematical programming approaches are outlined, discussing their advantages in ensuring solution rigor while highlighting inherent limitations in handling nonlinearity and large-scale problems. Subsequently, the global search capabilities and adaptability of mainstream metaheuristic and swarm intelligence algorithms are evaluated, with detailed analysis of AI- and machine learning-based methods enabling data-driven forecasting and adaptive control. A comparative assessment of centralized versus distributed optimization frameworks follows, highlighting trade-offs between scalability, data privacy, and real-time responsiveness. Representative applications in microgrid dispatch, energy storage, renewable grid integration, multi-energy coupling, and demand response are critically reviewed, with a focus on operational benefits and deployment challenges. Finally, key theoretical controversies are explored, including scalability barriers, privacy and security concerns, and issues related to model interpretability and engineering standardization. Looking ahead, the paper identifies future research trends: the convergence of edge and cloud computing, federated learning, cyber-physical collaborative optimization, and the pursuit of autonomous self-healing energy systems. This comprehensive analysis aims to provide researchers and practitioners with a structured reference, foster the exchange of ideas, and support the sustainable development of distributed energy management.