Data-driven solutions for microgrids energy management systems: A state-of-the-art survey on current trends and future directions

The rising global energy demand, depletion of fossil fuels, and growing environmental concerns have accelerated the integration of renewable energy sources into modern power systems. Microgrids offer an effective solution to enhance efficiency, reliability, and sustainability, but their operation is complicated by the intermittent and uncertain nature of renewable generation. Advanced energy management is therefore essential to achieve optimal performance. This paper presents a comprehensive review of energy management techniques, addressing alternating current, direct current, and hybrid network configurations, as well as operational objectives, constraints, and control structures, with a particular focus on data-driven approaches. Unlike earlier reviews, this study examines both model-based and data-driven strategies and includes a bibliometric analysis to identify research trends. The findings indicate that most existing approaches remain centralized and primarily target cost minimization, while ancillary services such as voltage regulation, frequency control, and cybersecurity remain underexplored. Although data-driven methods provide strong capabilities for forecasting and optimization, they face challenges related to data quality, overfitting, and computational cost. Future directions highlighted include neural networks and ensemble learning for accurate forecasting, reinforcement learning for adaptive real-time decision-making, data-driven model predictive control for operational scheduling, fuzzy logic for uncertainty handling, evolutionary algorithms for complex multi-objective optimization, and federated learning for decentralized and privacy-preserving coordination. By systematically mapping these methods and research gaps, this review provides practical insights for advancing sustainable energy management in microgrids.