【摘要】 随机几何偏差导致低压涡轮的气动性能呈现较强的分散性，这种影响在高负荷条件下进一步加剧。本文在不同负荷水平和攻角条件下，对几何偏差对涡轮叶片吸力侧边界层分离的影响进行了对比分析，并量化了几何偏差对失速裕度的不确定性影响规律。研究中，通过将叶片失速状态简化为二分类问题，并结合主动迁移学习策略，大幅提升了量化效率，有效降低了样本需求。结果表明，几何偏差导致涡轮裕度随机分散，显著缩小实际可用的高效工作范围，增大叶片失速风险，在实际气动设计中应予以充分重视。更多还原

【Abstract】 Random geometric deviations introduce significant variability in the aerodynamic performance of low-pressure turbines,which is further intensified under high loading conditions.In this study,the influence of geometric deviations on boundary layer separation on the suction side of turbine blades is compared across various loading levels and incidences.The consequent uncertainty impact on operating margins is quantified.To improve sample efficiency and reduce data requirements,the stall condition is simplified as a binary classification,and an active transfer learning strategy is employed.The results reveal that geometric deviations lead to substantial scatter in the operating margins of turbines,narrowing the available high-efficiency operating range and increasing the risk of stall.This uncertainty impact should therefore be carefully considered in the practical aerodynamic design of low-pressure turbines.更多还原