基于遗传算法的离心压缩机叶栅多点优化设计

MULTI-POINT OPTIMAL DESIGN FOR CASCADES OF CENTRIFUGAL COMPRESSOR BY GENETIC ALGORITHMS

离心压缩机元件的气动设计通常是根据一个给定工况点设计其几何形状。如此设计的元件在该设计工况点一般能保持较好的气动性态，但当实际运行工况改变时，性能往往会恶化。本文将机翼的多点设计思想引入离心压缩机叶栅的优化设计，并应用遗传算法对这类多目标优化问题进行求解，希望由此获得在运行工况间能较好折中的叶栅型线设计。以离心压缩机扩压器叶栅为例，在两个给定设计工况下采用所提出的方法求解其最佳叶片型线。数值结果显示，方法是有效和可行的。

The aerodynamic design for an element of centrifugal compressors is generally searching geometry of the element based on a given design condition. The element so designed usually has an expected aerodynamic performance near the design point, but becomes worse abruptly off the point. In the present paper, the concept of 'multi-point design' from approaches of airfoil design is introduced into the optimal design of cascades in centrifugal compressors. Genetic algorithms are used to solve that multi-point problem. It is expected that a trade-off optimal blade profile can be obtained, which makes a cascade having a good performance regarding to each given operation condition point. A diffuser cascade of a centrifugal compressor was chosen to be designed with the proposed approach, with its two operating points specified, and its blade profile to be optimized. The numerical results show that the proposed method can yield the trade-off optimal cascades between the two design points.