考虑不确定性的数值模拟模型确认方法  

Model validation method by considering uncertainty for numerical simulation

作　　者：任成坤 熊芬芬[2] 王元豪 李娜 姜波 郭志平 REN Chengkun;XIONG Fenfen;WANG Yuanhao;LI Na;JIANG Bo;GUO Zhiping(Southwest Technology and Engineering Research Institute,Chongqing 400039,China;School of Aerospace Engineering,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]西南技术工程研究所,重庆400039 [2]北京理工大学宇航学院,北京100081

出　　处：《西北工业大学学报》2023年第5期987-995,共9页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金(52175214)资助。

摘　　要：飞行器在其研发、生产和使用的整个寿命周期中都存在大量不确定性,这会导致数值模拟输出也具有不可忽视的不确定性,因此开展基于数值模拟技术的飞行器设计前必须进行模型确认。目前围绕数值模拟已开展了大量不确定性量化(uncertainty quantification,UQ)的研究工作,但是如何基于UQ的结果科学地进行模型参数修正,形成系统的数值模拟模型确认的闭环流程,构建高可信的数值模拟预测模型,还较少有研究报道。为此,提出一种考虑不确定性的数值模拟模型确认方法,发展一种基于优质小样本的模型参数修正策略,建立一套融合不确定性量化、全局灵敏度分析、参数修正的模型确认闭环流程,为构建高保真的数值模拟提供科学系统的思路,并通过翼型的仿真算例验证了该方法的有效性。There is a great amount of uncertainties in the whole life cycle of a flight vehicle,which can induce anon-negligible uncertainty in the numerical simulation output,then the model validation of the numerical simulation is the premise of design optimization.Lots of researches on the uncertainty quantification(UQ)of numerical simulation has been carried out.However,literature has rarely seen works about performing scientific model parameter correction by using the results of UQ and establishing a closed-loop procedure of numerical simulation model validation in a systematic way.Therefore,a numerical simulation model validation method by considering uncertainties is proposed,and a model parameter correction approach based on high-quality small samples is developed,and then a closed-loop model validation process composed by uncertainty quantification,global sensitivity analysis,and parameter correction strategy to provide a scientific and systematic procedure for constructing high-fidelity numerical simulations is established.The effectiveness and advantages of the model validation method is verified through airfoil simulation.

关 键 词：模型确认 不确定性量化 混沌多项式 深度学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] O242.1[自动化与计算机技术—控制科学与工程]