支持向量机和粒子群算法在结构优化中的应用研究  

Study and application of support vector machines and particle swarm algorithm in optimization design

作　　者：向国齐[1,2] 黄大贵[1] 严志坚[3] 

机构地区：[1]电子科技大学机电工程学院,成都610054 [2]攀枝花学院机电工程学院,四川攀枝花617000 [3]西南电子技术研究所,成都610036

出　　处：《计算机应用研究》2009年第6期2059-2061,共3页Application Research of Computers

基　　金：国家“973”重大基础研究资助项目(613580202)

摘　　要：针对实际复杂结构优化中计算量大的问题,提出将支持向量机代理模型和粒子群算法应用于工程优化设计。采用实验设计选取合适的样本,通过实验或数值仿真获得性能响应,利用支持向量机构建目标函数和约束的代理模型,重构原始的优化问题,采用粒子群优化算法对重构的优化模型进行寻优,从而得到最优解。以典型电子装备功分器的结构尺寸优化为例,采用拉丁方实验设计和高频电磁场仿真软件HFSS获取代理模型的训练样本,建立功分器模型的幅度比、相位差和驻波三个目标函数模型,并对该多目标优化问题进行寻优。结果表明该方法准确、高效,为结构优化设计提供了一种新的思路。Aiming at addressing the optimization design problems of computationally intensive simulation models with implicit objective performance functions, this paper proposed a framework based on the support vector machine and particle swarm optimization for structure optimization design. Selected appropriate design parameter samples by experimental design theories, obtained the response samples from the experiments or numerical simulations. Used the SVM method to establish the metamodels of the objective performance functions and other constraints, and reconstructed the original optimal problem. The reconstructed metamodels was solved by PSO algorithm. Adopted the structure optimization of the microwave power divider as an example to illustrate this methodology. Obtained the learning samples from uniform design theory and the high frequency electromagnetic field finite element analysis codes （HFSS）. Obtained three metamodels of the magnitude, phase and VSWR of the microwave power divider performance and solved the muhi-objective optimization problem. The resuhs show that this methodology is feasible and highly effective, and thus can be used in the optimum design of engineering fields.

关 键 词：支持向量机 粒子群优化算法 优化设计 代理模型 

分 类 号：H122[语言文字—汉语]