一种改进的协同优化过程研究  

Study on a kind of improved collaborative optimization

作　　者：杨维维[1] 赵勇[1] 陈小前[1] 姚雯 

机构地区：[1]国防科技大学航天与材料工程学院,湖南长沙410073

出　　处：《系统工程与电子技术》2007年第9期1564-1567,共4页Systems Engineering and Electronics

基　　金：国家自然科学基金资助课题(10302031)

摘　　要：在分析协同优化过程基本流程的基础上,针对现有过程中系统级协调算法计算量大、容易发散等缺点,提出了动态罚因子算法,该算法计算量小,优化效率高。采用正交试验设计选取初始设计向量,加快了优化收敛速度。用经典函数和齿轮减速箱优化问题对改进后的优化过程进行测试,算例结果验证了改进措施的有效性。Based on the analysis of the basic process of collaborative optimization（CO）, two improvements are presented. Firstly, to lessen the large computation burden and to avoid the divergence of the systemqevel optimization, a technique named as dynamic penalty factors is used, which has higher optimization efficiency, secondly, the orthogonal experimental design（OED） is used to choose the initial design vector to converge more quickly. Two examples are adopted to test the improved CO： a typical function and a gear reducer box design problem. The results show that the two improvements are very effective.

关 键 词：协同优化 动态罚因子 正交试验设计 

分 类 号：O224[理学—运筹学与控制论]