Experiment / simulation integrated shape optimization using variable fidelity Kriging model approach
Experiment / simulation integrated shape optimization using variable fidelity Kriging model approach
复制标题
DOI：
10.1299/jamdsm.2017jamdsm0053
复制
发表时间：
2017
影响因子：
0.9
通讯作者：
Wataru Yamazaki
中科院分区：
文献类型：
--
作者：
Wataru Yamazaki
对垂直轴风力涡轮机叶片翼型进行了试验/仿真一体化设计优化的尝试。在此集成设计优化中采用了变保真度Kriging代理模型方法，其中代理模型由高保真度和低保真度样本点信息构建。在本研究中，分别从实验和数值评估的方法定义了高保真度和低保真度的性能函数。在数值评估方法中，进行2D稳态计算流体动力学（CFD）模拟以评估翼型形状。在9个设计变量空间中建立了精确的变保真度代理模型，并在此基础上对不同翼型进行了优化设计。通过与仅通过CFD计算获得的另一种近似最优翼型的比较，讨论了本文集成设计优化的有效性和正确性。实验结果表明，与近似最优设计相比，该方法得到的最优设计的性能值提高了8%。更有效的形状优化可以实现包括低保真度的信息，其功能的趋势被用来构建准确的代理模型。
An attempt of experiment / simulation integrated design optimization is performed for the blade airfoil shape of a vertical axis wind turbine. A variable fidelity Kriging surrogate model approach is utilized in this integrated design optimization, in which a surrogate model is constructed from both high-fidelity and low-fidelity sample points information. In this research, the high and low fidelity performance functions are respectively defined from experimental and numerical evaluation methods. In the numerical evaluation method, 2D steady computational fluid dynamics (CFD) simulations are performed for the evaluation of an airfoil shape. An optimal airfoil shape is explored on the accurate variable fidelity surrogate model which is constructed in 9 design variables space representing various blade airfoil shapes. The validity and effectiveness of the present integrated design optimization are discussed by comparing with another approximate optimal airfoil shape which is obtained only by CFD computations. The optimal design obtained by the developed approach showed 8% larger performance value in the experimental evaluation compared with the approximate optimal design. More efficient shape optimization can be realized by including the low-fidelity information, whose functional trends are utilized to construct accurate surrogate models.