Traditional airfoil design is usually costly and time- consuming. As a complement to the traditional airfoil design, this paper combines CST parameterization and Gaussian Process Regression to efficiently predict aerodynamic performance and shape parameters of unknown airfoil. A group of four- digit NACA airfoils are selected. Their CST parameters,aerodynamic performance and pressure distribution are calculated as the experimental data. Using this data,fast airfoil aerodynamic performance evaluation and airfoil reverse design are implemented. The obtained results are compared with those obtained from NACA parameterization. Experimental results indicate that the CST- based prediction approach is accurate,fast and can be used to aid airfoil design in practice.%为弥补传统飞机翼型设计周期长、代价高的缺点,将CST翼型参数化方法,与机器学习中的高斯过程回归方法相结合,通过对已有的翼型数据的学习,实现对未知翼型气动性能或者外形数据的快速准确预测.选取一组NACA四位族翼型,获得其CST参数描述数据,并分别计算其在一定条件下的升力系数、阻力系数和压力分布数据.利用这些数据对高斯过程回归模型进行训练,实现了翼型的快速正设计以及反设计系统.并将实验结果与采用NACA翼型参数表示法得到的预测结果进行了对比.实验结果表明,基于CST参数化方法的翼型快速设计准确度高、速度快,具有很大的应用价值.
展开▼
