摘要
文章以iSIGHT设计软件为优化平台,集成Hypermesh和LS-DYNA,构建并实现了薄壁直梁件抗撞性多目标优化设计流程,在此基础上,运用拉丁超立方试验设计方法生成样本数据,采用响应面方法构建近似模型且其精度满足耐撞性设计要求;以比吸能SEA最大及碰撞峰值力Fmax最小为目标函数,采用多目标遗传算法NSGA-Ⅱ对响应面近似模型进行多目标优化设计,获取Pareto最优化解集,选取6个Pareto解进行有限元仿真验证。结果表明:近似模型方法能很好地取代实际仿真过程进行优化,且计算效率较高,基于6σ的多目标稳健性优化方法与确定性优化方法相比,稳健性优化结果提高了设计变量的可靠性和多目标函数的综合稳健性。
Choosing iSIGHT as an optimization platform ,the Hypermesh and LS-DYNA were integrat-ed to construct the procedure of the multi-objective optimization for the crashworthiness of thin-walled column .On this basis ,Latin hypercube design method was used to generate sample data ,and the re-sponse surface model（RSM） was used to fit the sample data and create approximate model .The accu-racy of RSM met the requirements of the crashworthiness design .A multi-objective optimization de-sign on RSM was implemented by multi-objective genetic algorithm NSGA-Ⅱ ,and the objectives were to maximize the SEA and to minimize the collision peak force Fmax .The Pareto optimal solution set was obtained ,and six Pareto solutions were selected and verified by finite element simulation .The re-sults showed that the approximate model could well replace the real process of simulation in optimiza-tion and could greatly improve the efficiency of optimization .Finally ,a multi-objective robust optimi-zation method based on 6σwas proposed .The robust optimal results improved the reliability of the de-sign variables and the comprehensive robustness of the multi-objective functions comparing to the cer-tain optimal results .
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2014年第2期145-149,共5页
Journal of Hefei University of Technology：Natural Science
作者简介
钱立军（1962一），男，安徽桐城人，博士，合肥工业大学教授，博士生导师．
