预测跨音速流场的区域分解与模型降阶方法  

Model Reduction with Domain Decomposition to Predict Transonic Flow

作　　者：王晨[1] 白俊强[1] 邱亚松[1] 李宇飞[1] 

机构地区：[1]西北工业大学航空学院,陕西西安710072

出　　处：《航空计算技术》2018年第1期21-25,共5页Aeronautical Computing Technique

基　　金：国家自然基金项目资助(11602199)

摘　　要：构建自适应降阶模型构造预测变几何、定常、跨音速翼型流场:利用本征正交分解(POD)方法建立降阶子空间;采用代理模型构建流场状态空间与降阶子空间的投影函数。由于跨音速翼型流场包含强烈的非线性现象(如激波等),使得降阶基模型难以准确捕捉流场特征,为提高模型在这类问题中的预测精度,在降阶模型中引入区域分解技术对激波区域及流场剧烈变化的区域单独建立降阶基模型。在跨音速翼型流场预测中,采用区域分解技术的模型降阶方法提高了原模型对跨音速翼型表面压力分布的预测精度,尤其是对激波处的模拟更加贴近全阶模型的计算结果。An adaptive reduced-basis model is built to predict parameterized steady transonic flow,where proper orthogonal decomposition( POD) is used to build the reduced subspace and surrogate method is in charge of projecting the state space onto reduced subspace. Because of the strong nonlinear effect induced such as shock wave in transonic flow,it is difficult for reduced models to accurately capture flow structures. In order to deal with this problem,domain decomposition technique is used,in which several other reduced models are built specifically just for region with shock wave or regions with large variation in state variables. In the transonic case,the reduced method with domain decomposition technique approximates the whole flow field more accurately,and models the region with shock wave better,though a little more time-consuming.

关 键 词：跨音速 本征正交分解 代理模型 区域分解 

分 类 号：V211[航空宇航科学与技术—航空宇航推进理论与工程]