中文摘要：

本文以自主发展的自适应遗传算法为基础,结合试验设计、神经网络响应面模型和三维流场数值模拟软件,完成了对设计空间的全局自动寻优。通过端壁型面的参数化控制,实现了压气机叶栅非轴对称端壁的气动数值优化,并分析了非轴对称端壁对叶栅气动性能的影响。结果表明：非轴对称端壁造型能够有效改变壁面的压力分布,从而改善角区流动,减少二次流动损失。相比于原型端壁叶栅,优化后总压损失系数降低了27.8%,静压升提高了19%。

英文摘要：

The aerodynamic numerical optimization method based on the adaptive genetic algorithm in house software is investigated.Adopting this method which is composed of design of experiment, neural network response model and numerical simulation software,the global optimization can be implemented automatically.The compressor cascade endwall has been optimized by the parameterization of endwall surface and the influence of the non-axisymmetric endwall on its aerodynamic performance is analyzed.The results show that the contouring endwall can effectively change the near endwall static pressure to improve the corner flow and decrease the loss of the secondary flow. Comparing with the datum cascade,the total pressure loss is reduced by 27.8%and the static pressure ratio is increased by 19%in the non-axisymmetric endwall.