A new optimization method of wind turbine airfoil performance based on Bessel equation and GABP artificial neural network

Abstract

One of the most important steps in designing wind turbines is to find airfoils with better performance. One of the major hurdles with parameterizing the entire airfoil shape, However, the large computational cost and complexity impose a major hurdle to analyze the airfoils in the optimization loop with parameterizing the entire airfoil shape. In order to solve this problem, GABP artificial neural network is used to optimize the design of airfoil. Bessel polynomial was used to simplify the airfoil curve to 8 pairs of coordinates. Then 1446 arrays were used as training set and 50 sets of data are used as test set. Finally, the ANN which can predict the lift coefficient and the maximum lift-drag ratio of airfoil is trained. The accuracy of the two parameters is 90%. In this paper, the characteristics of Bessel curve are used to train the neural network to optimize the airfoil. By adjusting the control points, the new airfoil can be created. It takes 168 s and has been adjusted 529 times, and the optimization target is successfully achieved. The method in this paper can provide new ideas for airfoil optimization and greatly reduce the optimization time. Furthermore, with the sufficient data input, the research can contribute to an efficient prediction and optimization on other airfoil performance.