支持向量机响应面在结构优化中的应用  

Response Surface Based on Support Vector Machine and its Application in Structural Optimization

作　　者：徐兴伟[1] 胡晓兵[1] 田昆[1] 李毅[1] XU Xing-wei;HU Xiao-bing;TIAN Kun;LI Yi(Sch. of Manufacturing Science and Engineering, Sichuan University, Sichuan Chengdu 610065, China)

机构地区：[1]四川大学制造科学与工程学院,四川成都610065

出　　处：《机械设计与制造》2018年第6期169-172,共4页Machinery Design & Manufacture

基　　金：四川省科技计划项目(2015GZ0032)

摘　　要：针对复杂工程优化问题,存在设计变量多、计算时间长、优化效率低且准确性难以判断等缺点。将支持向量机代理模型引入优化设计中。采用中心复合实验设计方法得到实验设计点,并采用支持向量机方法建立支持向量机响应面。对支持向量机响应面模型进行优化设计,得出优化结果。最后通过工程算例分析,建立对应的支持向量机响应面并对其进行优化设计,使用该方法优化后,其门式起重机主梁自重减轻了20%。为了验证其结果的准确性,采用有限元方法验证了优化后的结果,证明了其有效性。该方法对研究结构优化设计有着重要的参考及借鉴意义。In the design optimization process of complex engineering,the large amount of calculation and design variables、low computation efficiency are critical problem and it is difficult to evaluate the consequences. A response surface method based on support vector machines（SVM）was presented in the optimization design. Generate design point through central composite experimental design and then build support vector machines response surface model. Optimize design based on SVM response surface. It is concluded that the method have a significant effect in structure optimization design. Finally establish SVM response surface model for gantry crane. The weight of the structure are reduced by 25% after using SVM response surface model in optimization design. the accuracy of the results is verified by finite element method.it has an important reference structure optimization.

关 键 词：支持向量机 优化设计 代理模型 实验设计 

分 类 号：TH16[机械工程—机械制造及自动化] TH122