风洞MDOE的形式实验设计方法研究

Formal design of experiments method of MDOE in wind tunnel tests

MDOE风洞实验方法能够用相对于传统实验方法更少的吹风次数,获得更高精准度的数据。为了解决现有基于参数模型的MDOE方法获取较强非线性气动规律能力的不足,需要发展基于非参数模型的MDOE方法。本文对基于非参数模型的MDOE的形式实验设计方法进行研究。通过"虚拟"风洞实验的方法,对两种常用的"空间填充设计"——拉丁超立方设计和均匀设计应用于风洞实验的适用性进行对比,并在此基础上发展了几种对均匀设计的优化改进方法,可以进一步提高样本点设计质量,使其满足风洞实验的要求。研究表明:均匀设计较拉丁超立方设计更为稳健、均匀,更适合基于非参数模型的风洞MDOE方法;在均匀设计方法基础上,根据风洞实验的特点发展了优化方法,包括边界点补充、样本点密度调整和重复点设计,能够将已有的"先验信息"应用于实验设计中;所发展的形式实验设计方法所需的测量点要少于OFAT方法的测量点(如示例中所用的测量点数仅为OFAT方法的66.7%),且能够充分和准确地对较剧烈的非线性变化规律进行采样。本文对风洞MDOE的形式实验设计方法的研究结果,为后续发展基于非参数模型的风洞MDOE方法奠定了基础。

The existing MDOE methods based on the parametric model cannot meet the tunnel test requirements of advanced vehicle because of strong nonlinear aerodynamic behaviors.It is necessary to develop the MDOE method based on the non-parametric model instead.The formal design of experiments method of MDOE for wind tunnel tests is researched in this paper.The applicability of Latin Hypercube Sampling and Uniform Design in wind tunnel tests is compared through virtual experiments.And several optimizing methods based on Uniform Design are devoleped,which can improve the quality of design results to meet the requirement of wind tunnel tests.Studies show that,Uniform Design is more robust and uniform than Latin Hypercube Sampling for the MDOE method based on the non-parametric model,the optimization including boundary complement,density adjustment of sampling points and repetition points design can utilize the priori information to the design,the formal design of experiments method developed in the paper requires less sampling points than the OFAT,which are only 66.7%in the demonstration,and samples the intensive non-linear changing law adequately and exactly.The studies on the formal design of experiments method of MDOE in wind tunnel tests laid the foundation for further study of the MDOE method based on the non-parametric model.