Abstract: To improve the design quality of large-turning tandem cascades， an automatic multi-objective optimization system of tandem cascades was developed based on an improved particle swarm optimization algorithm， improved parallel multi-points infill criteria of Kriging model and physical programming. The system can achieve multi-objective optimization design of tandem cascade on design and off design incidence angle. In the multi-objective optimization system， to greatly reduce the computational scale of multi-objective optimization problems， the physical programming method is applied to transform the multi-objective optimization design problem into a single objective optimization problem based on the designer’s experience， and improved parallel multi-points infill criteria of Kriging model is used to generate multiple samples that are parallelly evaluated by numerical simulations per optimization cycle. The parallel multi-objective optimization system is validated by realizing the multi-objective of a large-turning tandem cascade. For the optimized tandem cascade， the static pressure ratio is higher and the total pressure loss coefficient is smaller at all incidence angle conditions. At inlet Mach number of 0.7， when incidence angle is -6° and 3°， the total pressure loss coefficient is respectively decreased by 21% and 35%， which proved that the parallel multi-objective optimization system has a good application value. Besides， the effects of five geometric parameters， such as camber ratio (TR)， chord ratio (CR)， axial overlap (AO)， percent pitch (PP) and incidence angle of the rear airfoil (KBB) on the design and off design performance of tandem cascades were studied in detail. Results indicate that tandem cascades with a high PP (about 0.9) and a negative KBB (about -6°) can realize good flow performances on design incidence angle. And a smaller load of front airfoil can improve the flow characteristics of tandem cascades off design incidence angle and increase the stable operation range.

Key words: Tandem cascades；Multi-objective optimization；Parallel multi-points infill criteria；Physical programming

摘要： 为了改善高负荷串列叶栅的设计质量，基于改进粒子群算法、Kriging模型的改进并行多点采样策略、物理规划方法三个模块，建立了一套多目标优化设计系统，该系统可以快速实现串列叶栅设计攻角和非设计攻角的多目标优化设计。为了减少多目标优化的计算量，该优化设计系统采用了物理规划方法将多目标优化设计问题转化为考虑设计者经验的单目标优化问题，并基于Kriging模型的改进并行多点采样准则实现了在一次迭代过程并行评价多个样本点的并行优化方法。应用该系统实现了一高负荷串列叶栅的多目标优化，优化后的串列叶栅在全攻角下的总压损失系数减小，静压升增加，在进口马赫数0.7的条件下，在攻角分别为-6°和3°时，总压损失分别降低21%和35%，证明了本文设计的多目标优化系统具有很好的实际应用价值。基于优化设计结果，分析了串列叶栅的5个造型参数:弯角比(TR)、弦长比(CR)、后排近似攻角(KBB)、轴向重叠度(AO)和节距比例(PP)对串列叶栅设计攻角和非设计攻角性能的影响，研究发现大的PP(约为0.9)和负的KBB(约为-6°)有助于串列叶栅实现较优的设计攻角性能，减小串列叶栅前排叶型的负荷，可以改善串列叶栅非设计攻角的性能和扩宽叶栅的稳定工作范围。

关键词: 串列叶栅；多目标优化；并行多点采样策略；物理规划