飞机多学科设计的分级优化方法

Exploration of Aircraft Multidisciplinary Design Optimization with Hierarchical Decomposition

飞机设计是十分复杂的过程 ,它需要多个学科的综合及许多部门的合作。基于层次型分解的多学科设计优化 (MDO)算法就是解决这类复杂工程系统优化的有效方法。它将大系统分解为许多层 ,每层又有许多子系统 ,而各个子系统的计算规模较小 ,同时又能保留各层次间的耦合关系。通过三梁元刚架结构的设计 ,证实了本文算法的有效性。

We are among the few in China to engage in hierarchical decomposition research. In our MDO (multidisciplinary design optimization) algorithm, we decompose the design optimization problem into three hierarchical levels. We begin repeated iterations from top in the hierarchy. First we make any reasonable distribution of resources at the top and let the top hierarchical level allocate resources to medium hierarchical level and we again make reasonable distributions at the medium level and let the medium level allocate resources to the bottom level. The real optimization begins at the bottom level and from then on the down-up and up-down iterations are all really optimizations based on preceding results of optimization. The iterations are repeated until a satisfactory optimization is attained. In this way, the complicated interactions among disciplines are paid attention to without undue time consumption. We present the optimization of a three-beamed framework to demonstrate the algorithm and its effectiveness.