自适应协同优化方法研究

针对一致性约束条件造成协同优化方法计算困难的问题,研究提出了一种自适应协同优化方法.首先将设计变量空间划分为三类区域,分析了每个区域内一致性约束松弛量与系统可行域和学科一致性的关系,并在此基础上提出了三种一致性约束松弛量的确定方法.然后基于上述方法,给出了自适应协同优化的算法流程,对比分析了相对于标准协同优化算法流程的改进之处.最后采用自适应协同优化方法对两个典型MDO算例进行优化,证明了该方法具有较好的鲁棒性和收敛速度.

Research on adaptive collaborative optimization method

Aiming at the computational difficulties of collaborative optimization caused by consistency constraints,an adaptive collaborative optimization method is presented.Firstly,design variables space is split into three kinds of regions,then the relations among the relaxed tolerances of consistency constraints,system feasibility and subsystem consistency are analyzed within each region.Based on that analysis,three methods of relaxed tolerances to calculate are given.Secondly,the algorithm flow of the adaptive collaborative optimization method is proposed,and the improvement is analyzed comparing with the standard collaborative optimization method.Finally,two classic MDO test suites are optimized by the proposed adaptive collaborative optimization method,and the result shows its advantage on robustness and convergence rate.