一种基于协同近似的多学科设计优化方法

为避免求解复杂工程系统多学科设计优化问题时需要反复调用复杂耗时的多学科分析和进行繁琐的灵敏度计算，提出了一种基于协同近似（CMSO）的多学科设计优化方法。首先通过一种协同模型来筛选出更能反映多学科问题本身属性的样本点，通过协同模型来保证系统的多学科一致性；然后通过这些样本点分别构建自适应代理模型，并进行代理模型的验证和确认；最后选择最佳的代理模型来构建多学科优化模型并使用序列二次规划法进行优化求解。通过一个数学算例和圆柱螺旋压缩弹簧设计案例验证了CMSO法的可行性和有效性，且与单学科可行法的比较结果表明了CMSO法的高效性。

CMSO for Multidisciplinary Design Optimization Method

In order to avoid complex time-consuming multidisciplinary analysis and cumbersome sensitivity calculation in solving complex engineering multidisciplinary design optimization problems, a CMSO method was proposed for multidisciplinary design optimization. Firstly, a collaboration model was used to filter sample points which might reflect the characteristics of the multidisciplinary design optimization problem better and to maintain the multidisciplinary consistency of the system. Then, adaptive surrogate models were constructed, verified and confirmed through the selected sample points. Then the best surrogate model was selected to approximate the multidisciplinary design optimization model and the sequence quadratic program method was used for optimization solution. The feasibility and effectiveness of the CSMO method were verified by a mathematical example and a cylindrical spiral compression spring design case, and the high efficiency of CMSO method was demonstrated by comparison with individual discipline feasible.