Metaheuristic Algorithms for the Optimal Control of Microgrids: Current Status Analysis and Prospects

Microgrids integrate distributed generation, energy storage systems, and diverse loads into a localized energy network capable of operating in both grid-connected and islanded modes. With the rapid growth of renewable energy penetration, the intermittency and volatility of photovoltaic and wind generation, combined with heterogeneous resources and time-varying load demands, make microgrid control and operation increasingly challenging. Modern microgrids are typically organized in a hierarchical control architecture. Primary control enables fast inverter-level regulation for voltage, current, and frequency stabilization and supports basic power sharing. Secondary control mitigates the steady-state deviations introduced by droop mechanisms and supports smooth transitions between grid-connected and islanded modes. Tertiary control performs system-level optimization in the energy management layer by coordinating power references, droop set points, and allocation coefficients to achieve global performance in terms of efficiency, economy, stability, and renewable accommodation. However, converting system-level goals into coordinated and executable commands for distributed units remains difficult because of component coupling and uncertainties arising from renewables, loads, and market conditions. In practice, microgrid optimization is often nonlinear and nonconvex and may become mixed integer and multi-objective when unit commitment, discrete actions, network losses, and multiple indices are considered. The objectives typically include minimizing operating costs and emissions, maximizing renewable utilization, and improving reliability, power quality, and resilience under constraints such as power balance, device limits, ramping requirements, and operational security. Conventional optimization methods can work well under simplified models; however, their effectiveness may degrade under nonlinear losses, nonconvex feasible regions, dynamic electricity prices, weather-driven uncertainty, and high-dimensional decision spaces, leading to heavy computation, limited adaptability, and local optimality. Therefore, metaheuristic optimization algorithms have attracted attention for microgrid optimal control owing to their strong global search capability and flexibility under uncertainty. This review summarizes metaheuristic-based microgrid optimization for four representative scenarios. Energy management addresses multi-objective and multi-time scale coordination and may be extended to multi-energy carriers. Economic dispatch focuses on short-term real-time scheduling under renewable energy variability, demand response, and market interactions. Resilience enhancement uses metaheuristics for controller tuning and event-triggered strategies or reconfigurations to strengthen voltage and frequency stabilities and accelerate post-disturbance recoveries in islanded operations. Anomaly detection integrates metaheuristic-assisted learning with hybrid diagnosis to enable rapid and accurate fault identification and localization in complex data streams under limited samples. The key challenges include data quality and operability, algorithmic issues such as premature convergence and parameter sensitivity, and physical bottlenecks in computation and communication. Future research directions include robust data preprocessing and scenario generation, hybridization with model-based solvers, and distributed parallel architectures such as edge cloud collaboration and hardware acceleration to meet real-time requirements and support reliable engineering deployment.