基于POD和代理模型的热气防冰性能预测方法

Hot air anti-icing performance estimation method based on POD and surrogate model

大型客机机翼、短舱多采用热气防冰作为主要防冰策略。为了缩短热气防冰系统优化设计周期,提出了基于本征正交分解和代理模型的防冰性能快速预测方法。采用本征正交分解对数值仿真积累的温度和溢流水快照进行特征分析,选取包含绝大部分样本特征的基模态线性拟合所有快照;基于支持向量机回归方法建立笛形管结构参数与线性拟合系数间的代理模型,实现对热气防冰蒙皮外表面温度分布和溢流水分布的快速预测。针对三维缝翼笛形管热气防冰系统开展的验证表明:该方法对防冰表面温度分布的预测效果较好,并能够较为准确地预测水滴撞击区域内的溢流水分布;建立的预测方法计算成本较数值计算方法大幅降低,对于热气防冰系统优化设计工作具有重要意义。

Most large civil aircraft use hot air anti-icing systems as anti-icing strategies for airfoils and nacelles.A novel estimation method for hot air anti-icing system performance based on Proper Orthogonal Decomposition(POD)and surrogate models is proposed to shorten the design cycle of hot air anti-icing system.POD is adopted for data compression and characteristics extraction for the anti-icing performance snapshot matrix obtained by numerical calculation,and a lower-dimensional approximation for the snapshot matrix is derived from the projection subspace consisting of a set of basis modes.Support vector regression method is used to construct the surrogate models between the fitting coefficients of basis modes and the piccolo tube geometric parameters.The validation of the estimation method on a three-dimensional slat piccolo tube hot air anti-icing system shows that this method has a good surface temperature prediction and can accurately predict the runback water distribution within the droplet impingement area.The time consumption of the established estimation method reduces significantly compared to the numerical simulation method,which is of great significance for the hot air anti-icing system optimal design.