摘要
多学科设计优化在实际应用中,几乎都采用了可变复杂度方法(variable-complexity method,VCM)来平衡计算成本和计算精度。然而,VCM通常需要计算大量的初始样本点以保证精度。为降低VCM对初始样本点的依赖,文中提出了一种基于响应面更新的可变复杂度方法。以桁架式Spar平台的多学科设计优化为例,首先根据试验设计选取样本点,采用数值方法计算平台的各项性能,建立了响应面模型;应用协同优化算法求解,并通过响应面的不断更新提高精度,在此基础上应用可变复杂度方法校正响应面分析结果。计算结果表明:基于响应面更新的可变复杂度方法可在较少初始样本点情况下,获得较高精度的计算结果。
In the application of multidisciplinary design optimization 0VIDO）, variable-complexity method （VCM） was widely used to balance the computational etficiency and accuracy. However, VCM generally requires large number of calculations for initial sampling points to ensure precision. In order to overcome such a deficiency, an improved VCM strategy utilizing adaptive response surface method is proposed. The strategy is illustrated in the application of MDO to Truss Spar concept design. In this study, multidisciplinary numerical analyses are conducted for initial sampling points to build the response surface models （RSMs）. A collaborative optimization is used as MDO algorithm and RSMs are sequentially updated until convergence criterion is met. Finally, VCM is adapted to correct RSMs by using the results of high-fidelity analysis tool. The results show that the improved VCM strategy can attain high accuracy with a small number of initial sampling points and is an effective way to minimize the comnutationai cost.
出处
《船舶力学》
EI
北大核心
2010年第7期771-781,共11页
Journal of Ship Mechanics
关键词
多学科设计优化
可变复杂度方法
响应面更新方法
桁架Spar平台
multidisciplinary design optimization
variable-complexity method
adaptive response surface method
Truss Spar
