Airfoil Robust Optimization Based on Convolutional Neural Network and Polynomial Chaos Method

Abstract

In conventional airfoil optimization design method, the aerodynamic performance of the optimal airfoil can deteriorate at the non-design point, so it is necessary to study the airfoil robust optimization method.An airfoil robustness design method based on convolutional neural network and polynomial chaos method is proposed.Firstly, an aerodynamic force prediction method based on convolutional neural network model is established.And then, the polynomial chaos method is used to perform the uncertainty quantification for mach number and attack angle, a design system of airfoil robust aerodynamic optimization is constructed.Finally, the optimization design verification is carried out for the aerodynamic optimization design problem, in which the RAE2822 airfoil is taken as the basis airfoil.The results show that the robustness design method of the airfoil proposed in this paper is feasible, the optimized airfoil aerodynamic performance and robust aerodynamic optimization design efficiency are improved in a wide design range.

Keywords

• convolutional neural network
• polynomial chaos method
• airfoil aerodynamic optimization
• robust optimization
• uncertainty quantification