飞机多学科设计优化中的并行多目标子空间优化框架

Concurrent Multi-objective Subspace Optimization Framework in Aircraft Multidisciplinary Design Optimization

针对现有的并行子空间优化框架存在的各子空间仅能有一个优化目标等局限,提出了并行多目标子空间优化框架。该框架将各个学科的优化问题由以往的单目标优化改进为多目标优化,使得每个子空间可以分配到多个优化目标、各子空间的设计变量可以重叠,并且在一次优化中就可获得优化问题的Pareto前沿。介绍了新框架的基本思想和流程,并且阐述了新的学科解耦、子空间的并行优化和系统级的设计变量综合方法。以一个飞机总体设计问题为例,考虑气动、隐身与控制学科,对翼面几何参数和控制律参数进行了优化设计,验证了并行多目标子空间优化框架的有效性和相对于已有方法的优势。

Existing concurrent subspace optimization frameworks are generally just applicable to problems with only one objective in each subspace. To overcome such limitations, concurrent multi-objective subspaee optimization （CMOSSO） is proposed which can improve the optimization process of each discipline from single-objective optimization to multi-objective optimization. By this means, more than one objective can be set in each subspace, design variables in different subspaces can overlap, and the Pareto front of the problem can be obtained in one run of optimization. The basic ideas and process of CMOSSO are introduced; new methods for disciplinary decomposition, concurrent subspace optimization, and synthesis of design variables at the system level are depicted. An aircraft conceptual design problem is taken as an example, which takes into account. Three disciplines including aerodynamics, stealth,and control, and the geometric parameters of wing and empennages as well as control rule parameters are optimized. Through this case, the effectiveness and advantage of CMOSSO over existing methods are validated.