基于相对差商与序列两级算法的离散变量结构优化研究  

Research of optimization design of structures with discrete variables based on RDQA and sequential two-level algorithm

作　　者：张伟[1] 邢菲[2] 顾亮[3] 

机构地区：[1]开封大学实验实训处,开封475004 [2]开封大学机电工程学院,开封475004 [3]大连理工大学精密与特种加工教育部重点实验室,大连116024

出　　处：《机械设计与制造》2009年第7期160-162,共3页Machinery Design & Manufacture

基　　金：高等学校博士学科点专项科研基金(20050141010)

摘　　要：针对离散变量结构优化设计中存在的NP问题,提出了相对差商-序列两级混合优化设计思想,建立了混合优化数学模型。在可行集外由目标函数最小点出发,沿目标函数增加最小、约束函数降低最多的方向逐步搜索,通过整体与局部两级约束划分,提高了收敛速度与全局搜索能力。比较结果表明,该方法在解决一类目标函数与约束函数具有单调性质的离散变量优化问题中,具有很好的可行性。As the NP problem in optimization design of structures with discrete variables ,the idea of mixed optimization with relative difference quotient algorithm（ RDQA ） and sequential Two-Level algorithm had been put forward. The mixed optimization model had also been established. This method starts from the minimum point of objective function out of feasible set and advancing along the direction of minimum increase of objective function and maximum decrease of constrained functions,the rate and the ability to overall situation research had been increased dividing unitary and partial limitation. It shows that it is more efficient for the optimization which its objective and restrict function are flat.

关 键 词：离散变量 相对差商 序列两级 混合优化设计 

分 类 号：TH16[机械工程—机械制造及自动化] O342[理学—固体力学]