Wind turbine design optimization faces significant challenges due to its multi-objective, nonlinear dynamics influenced by uncertain environmental parameters like wind speed and air density, coupled with complex structural constraints. This study proposes a novel probabilistic framework combining active learning Kriging (ALK) with a hybrid Grey Wolf Optimizer (GWO) and Arithmetic Optimization Algorithm (AOA) to simultaneously minimize the Cost of Energy (COE) and the Annual Energy Production (AEP) variability under uncertainty. After rigorous validation through benchmark problems, the method was applied to wind turbine (WT) design under both deterministic and probabilistic conditions. Results demonstrated 15-20% lower AEP variations and 30% reduced failure probability in probabilistic designs compared to deterministic approaches, while deterministic solutions achieved 5-8% higher AEP at marginally lower COE. Sensitivity analysis revealed that rated power is the most influential parameter, affecting COE and AEP more significantly than rotor radius or hub height. The framework provides practitioners with a robust tool for optimizing onshore WT configurations under real-world uncertainties, particularly valuable for wind farm planning in variable climate conditions.
Fang et al. (Wed,) studied this question.