【摘要】 风扇/增压级作为航空发动机的核心部件，其性能直接影响发动机整机的效率和推力输出。为满足现代航空风扇/增压级设计方法的更高需求，需要对传统的流线曲率(SLC)法改进和优化，采用神经网络对传统经验模型进行替代，通过人为构造叶栅流场数据集对神经网络模型进行训练，并在传统流线曲率法算法上耦合该模型以形成计算程序。利用该程序对风扇/增压级进行性能预测，并将计算结果和传统流线曲率法与3维数值模拟结果进行对比。结果表明：经改进后的流线曲率法，其预测结果相较于传统算法，精度平均提高约2%，从而实现了对流场的准确预测。更多还原

【Abstract】 As a core component of aeroengine,the fan/booster directly affects the overall efficiency and thrust output of the aeroengine. To meet the increasing demands of modern fan/booster design methods, improvements and optimizations of the traditional streamline curvature(SLC) method are required. In this study,neural networks are employed to replace conventional empirical models. A blade-row flow-field dataset is artificially constructed to train the neural-network model,which is then coupled with the traditional SLC algorithm to form a computational program. This program is applied to predict the performance of a specific fan/booster,and the results are compared with those obtained by the traditional SLC method and by three-dimensional numerical simulations. The results show that the improved SLC method yields prediction accuracy about 2% higher on average than those of the traditional algorithm. Thereby a more accurate flowfield prediction is enabled.更多还原