Metaheuristic Optimization and Artificial Intelligence in Wind Energy Systems for Sustainable Environment: Forecasting, Control, Design, and Diagnostics—A Critical Review

Wind power is the largest renewable energy source used at the sustainable utility scale. However, wind turbine design, siting, forecasting, control, and maintenance face real challenges, mainly because wind speed is variable and poorly predictable. Over the last ten years, researchers have applied both artificial intelligence (AI) models and metaheuristic optimization algorithms across all phases of the wind energy value chain. This review examines these two methodological approaches across four application fields, with particular attention to evaluation practice, validation level, and implications for sustainability. First, we review wind speed and power forecasting, from statistical baselines to machine learning and deep learning architectures, including models that rely on metaheuristics to tune their hyperparameters and decomposition stages. Second, we examine control applications, focusing on metaheuristic tuning of generator and converter controllers (mainly rotor-side control of doubly-fed induction generators), pitch and torque control, and maximum power point tracking (MPPT). Third, we discuss design-level optimization, covering blade aerodynamic design and wind farm layout under wake effects. Fourth, we examine machine-learning-based supervisory control and data acquisition (SCADA) monitoring and fault diagnosis, including normal-behavior modelling and anomaly detection. Finally, as floating offshore wind turbines (FOWTs) are a cross-cutting emerging field, with coupled hydro-aero-elastic problems in control co-design and AI-based monitoring, the paper proposes a research agenda for the next sustainable generation of onshore and offshore wind systems, focused on standardized open benchmarks, uncertainty-aware forecasting, physics-informed learning, and hardware-validated optimization.