并行GSE方法及其在耦合系统优化中的应用

指出了并行全局灵敏度方程（global sensitivity equation, GSE）方法的优势，定性分析了并行GSE方法的计算性能。在ModelCenter框架下开发了通用并行GSE驱动组件(general parallel GSE driver, GPGDRV)，并与优化算法相结合,基于COM技术开发了通用并行GSE优化器(general parallel GSE optimizer, GPGOPT)，进而将并行GSE方法用于耦合系统优化。数值算例表明，GSE方法可以显著提高耦合系统灵敏度的求解效率。将GPGDRV和GPGOPT用于某假想的教练机方案优化设计，并与有限差分法进行计算性能定量比较，定量比较结果与定性分析相符。研究还表明，GSE方法极大地提高了耦合系统优化的效率，使优化时间缩短约76%，并行GSE方法可进一步缩短优化时间。GPGDRV和GPGOPT对于耦合系统的分析和优化具有实用性。

Parallel GSE method and its application in optimization for coupled systems

The advantages of parallel global sensitivity equation (GSE) method are pointed out, the computational performance of the parallel GSE method is also qualitatively analyzed. The general parallel GSE driver (GPGDRV) is developed based on ModelCenter, and then the general parallel GSE optimizer (GPGOPT) is created through COM technology associated with an optimization algorithm. In this manner, the parallel GSE is applied in optimization for internally coupled systems. Numerical study case illustrates that the efficiency to solve the sensitivity of coupled systems is improved evidently through the GSE method. A conceptual optimization of an assumed training airplane is carried out with GPGDRV and GPGOPT, a quantificational comparison of computational performance with the finite difference method is conducted as well. The comparative results agree well with the qualitative analysis. Moreover, the study results demonstrate that computational efficiency is improved  dramatically by using the GSE method, which makes the optimization time decrease by about 76%. The optimization time, furthermore, can be still reduced ulteriorly through the parallel GSE method. Additionally, GPGDRV and GPGOPT are practicable for the analysis and optimization of coupled systems.