This paper presents a hydrodynamic performance prediction method based on Joukowski transformation and Gaussian process regression for rapid airfoil design of small-scale tidal stream turbines operating in low-velocity conditions. A set of Joukowski airfoils were selected as initial profiles and fitted using CST parameterization method. The lift and drag forces under low-velocity conditions were obtained through Fluent numerical simulation. A hydrodynamic performance prediction model was established by combining lift
drag
CST parameters with Gaussian process regression based on machine learning. Experimental results show that the average error rates of lift and drag prediction models are below 0.2%
RASMUSSEN C E, WILLIAMS C K I. Gaussian Processes for machine learning (adaptive computation and machine learning)[M]. Cambrige, MA: The MIT Press, 2005.