分布式进化算法及其在翼型气动反设计中的应用

构造了一种新型基于基因算法与博弈论的并行分级多目标优化方法，并应用于多段翼型气动反设计。此方法基于二进制编码的基因算法和博弈论，优化变量被分配给不同的博弈者，因而总体优化问题转变为分裂空间中的局部优化问题。文中给出了一个多段翼型形状，位置可压位流的反设计问题的求解算例，引入了基于非结构网格的分级结构。与传统基因算法数值算例的对比表明了本文构造的并行分级算法具有较高的计算效率，可广泛应用于多目标优化问题。

Distributed evolutionary algorithms for inverse design problems of airfoil in aerodynamics

New parallel hierarchical multi-objective optimization approaches based on Genetic Algorithms (GAs) with Nash scenarios of Game Theory (GT) are investigated for solving inverse multi-element airfoil design problems in aerodynamics on distributed parallel environments. A multi-objective optimization methodology presented here relies on binary coded GAs and coupled with GT. The design variables of such optimization problems are split among several players, the global multi-criterion optimization problem being replaced by several sub optimizations operating in the decomposed search space. A shape/position reconstruction problem (inverse problem) for a multi-element airfoil in compressible potential flow is solved using Parallel Hierarchical GAs coupled a Nash game with a hierarchy based on unstructured meshes. Numerical results, compared with sequential algorithms, show that parallel hierarchical GAs combined with Nash strategy are more efficient and robust than simple GAs and this method could be used with high efficiency for complex multi-criteria optimization problems in aerodynamics.