An Intelligent Parameterization Method of Large Wind Turbine Airfoil Considering Geometric Constraints

Conventional parametric methods fail to enforce geometric constraints during airfoil sampling, so the relative thickness of the sampled airfoils is hard to control and spans a wide range, and the larger design space leads to lower design efficiency. To meet the growing demand for thick airfoils in large-scale windturbine blades and to support their efficient development, this study proposes an intelligent parameterization method for airfoils based on Generative Adversarial Networks (GAN) and CST (Class-Shape Transformation). Deep learning techniques are used to learn the intrinsic geometrical laws of the airfoil shape and autonomously generate a new airfoil shape with controllable thickness, high smoothness and high accuracy. Under the condition of satisfying thickness constraints, the method significantly compresses the design space, thereby enhancing airfoil optimization efficiency.