A Supercritical Airfoil Design Based Onmulti-output Surrogate Model

The airfoil is the main component of the aircraft to generate lift and drag, which has a significant effect on the performance of the aircraft. Airfoil aerodynamic shape affects the cruising speed, takeoff and landing performance, handling qualities, stall properties etc. Therefore, the optimization and design of airfoil has become an essential part in the design of airfoil. Traditional Airfoil design methods, such as wind tunnel test and CFD technology, seriously restricted the efficient design of airfoil due to the high cost and long design cycle.This paper proposes an airfoil optimization method based on the combination of multiple output Gaussian process regression model and genetic algorithm and makes an intensive study of the aerodynamic shape optimization design of supercritical airfoil. First, geometry disturbance is added to the initial airfoil to produce a reasonable space of airfoil design. Then, the Latin hypercube method is used to construct a series sample points in the design space and the CST parameterization method is used to define the airfoil geometry shape. As response values, the corresponding lift coefficient and drag coefficient of airfoil, are obtained by grid generation tool(Gridgen) and CFD calculation in order to establish the initial MOGP proxy model. The model simulates the function relation between the input and output. In the given airfoil space, the optimal airfoil can be obtained through genetic algorithm. At the same time, the aerodynamic performance of airfoil can be predicted by MOGP model. The design goal is to minimize the drag coefficient with consideration of some constrained conditions such as area, lift and so on. The results of single point and multi-point optimization tests show that the proposed method can effectively reduce the drag coefficient and optimize the pressure distribution of airfoil while the area and lift coefficient remain constant. That is to say, the optimization design method based on MOGP model can achieve the goal of the optimization design.