基于降阶模型的翼型结冰冰形预测方法

翼型结冰冰形的数值模拟预测通常比较复杂耗时，为了更加快速准确地预测冰形以减少计算资源消耗，建立了基于本征正交分解（POD）和Kriging模型的冰形快速预测方法。利用CFD数值模拟结果来构建样本空间，以飞行迎角为例详述了降阶模型的冰形预测的实现手段，并结合试验设计方法，完成了多参数的结冰冰形快速预测，同时研究了先进的Blind-Kriging模型的相关方法以及对于预测结果的改进。结果表明，降阶模型预测翼型结冰冰形与CFD数值模拟结果吻合较好，表明降阶模型可以快速、精确地应用于翼型结冰冰形预测。 

Ice shape prediction method of aero-icing based on reduced order model

The numerical simulation prediction of aero-icing ice shape is usually complicated and time-consuming. In order to predict ice shape more quickly and accurately and to reduce computational resource consumption, a quick ice shape prediction method based on proper orthogonal decomposition and Kriging model is proposed in this paper. Using a database of high-fidelity CFD numerical simulations to build sample space, in view of the change of attack angle, the procedure for predicting the ice shape by reduced order model is introduced. With the data sampling method of parameter space, multiparameter prediction of the ice shape is achieved. Meanwhile, the related methods of the advanced Blind-Kriging model and the improvement of the prediction results are studied. The result shows that the prediction results of the airfoil icing shape using the reduced order model agree well with the CFD numerical simulation results. The conclusion is made that the reduced order model is a quick and accurate approach for predicting the ice shapes of airfoil icing.