基于机器学习的压气机叶型优化设计

The performance requirements of modern compressors are increasingly stringent, and the optimization of blade is very important. In recent years, genetic algorithm has been widely used in blade optimization design, while, the traditional genetic algorithm is too time-consuming. In order to solve this problem, this paper proposes a flow field reconstruction method based on principal component analysis and artificial neural network. Additionally, an optimization process based on this method to quickly evaluate the performance of blade is proposed. Numerical results show that the error between the predicted static pressure ratio based on the optimization process and the CFD calculation value is less than 0.1%, which can effectively achieve the optimization target. And the static pressure ratio of this optimization result is increased by 14.7% compared with the original blade, thus, the effects of the optimization is considerable.