Toward improved siting of wind–solar hybrid farms: A novel framework integrating multi-criteria decision making, machine learning–driven feature selection, and spatial clustering

Creative Commons License

Uguz Y. F., Bilgili A., Uzar A. M.

Applied Energy, cilt.411, ss.1-23, 2026 (Scopus)

Özet

Ensuring secure, affordable, and sustainable energy supply is a growing global challenge as rising demand and supply insecurity accelerate the transition to renewable energy. Hybrid wind–solar systems offer an effective solution by exploiting spatial and temporal resource complementarity, but their deployment critically depends on reliable site selection. Most multi-criteria decision analysis studies rely on a single weighting method, apply limited criteria evaluation, lack statistical validation, and rarely identify spatially coherent, investment-ready zones. This study addresses these gaps by proposing an integrated geographic decision-making framework for hybrid wind–solar siting that combines two expert-based weighting methods, Analytical Hierarchy Process (AHP) and Best-Worst Method (BWM) with an objective, data-driven weighting method, CRiteria Importance Through Intercriteria Correlation (CRITIC) and machine learning–assisted criterion screening. The framework was applied to Izmir Province, Türkiye. Suitability maps generated under alternative weighting methods were compared using spatially adjusted statistical tests, clustered into contiguous high-suitability zones using hierarchical density-based spatial clustering (HDBSCAN), converted into candidate-site polygons, and prioritized using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and the Multi-Attributive Border Approximation Area Comparison (MABAC) method. The results revealed pronounced differences across weighting methods. Highly or very highly suitable areas accounted for 33.27% (AHP) and 27.75% (BWM) of candidate sites, compared with 58.39% under CRITIC. Validation using 703 existing solar photovoltaic and wind turbine installations showed agreement rates of 77.81%, 72.83%, and 83.65%, respectively. The resulting outputs provide actionable decision-support information for regional planning and the identification of investment-ready hybrid wind–solar zones.