论文部分内容阅读
提出一种新的多学科设计优化方法,即子空间分解与淘汰优化方法。该方法通过子空间的分解和淘汰,提高剩余子空间的近似模型精度,基于子空间近似模型优化获取最优解。首先,基于设计空间近似模型获取最优解,如果近似模型达到满意精度,则终止优化;否则将设计空间分解为多个子空间。然后,各子空间基于近似模型优化,如果子空间没有可能获得优于当前最优解的最优解,则淘汰;如果子空间近似模型的精度达到满意精度,则子空间不再分解;如果子空间没有获得满意精度但有可能获得更优解,则将子空间分解为更小的子空间。该方法的优化计算时间与设计变量维数和设计空间大小密切相关。算例研究表明该优化方法在计算时间和全局优化解方面具有良好性能。
A new multidisciplinary design optimization method is proposed, which is subspace decomposition and elimination optimization methods. This method improves the accuracy of the approximate model of the remaining subspaces through the decomposition and elimination of subspaces, and obtains the optimal solution based on the subspace approximation model. First, the optimal solution is obtained based on the approximate model of the design space, and the optimization is terminated if the approximate model reaches the satisfactory accuracy; otherwise, the design space is decomposed into multiple subspaces. Then, the subspaces are optimized based on the approximate model. If the subspaces are not likely to obtain the optimal solution better than the current optimal solution, the subspaces are eliminated. If the subspace approximation model achieves satisfactory accuracy, the subspaces are no longer decomposed. Space does not get satisfactory accuracy but it is possible to obtain better solution, the subspace is decomposed into smaller subspaces. The optimization calculation time of this method is closely related to the dimension of design variables and the size of design space. The case study shows that the optimization method has good performance in computing time and global optimization solution.