基于BP神经网络的侧碰多目标优化设计

中国机械工程 ›› 2012, Vol. 23 ›› Issue (17): 2122-2127.

基于BP神经网络的侧碰多目标优化设计

周利辉;成艾国;陈涛;赵敏;周泽

湖南大学汽车车身先进设计制造国家重点实验室,长沙,410082

出版日期:2012-09-10 发布日期:2012-09-12
基金资助:
湖南大学汽车车身先进设计制造国家重点实验室自主研究课题资助项目(60870002);噪声振动和安全技术国家重点汽车实验室2011年度开放基金资助项目(NVHSKL-201103)

#br# Multi-objective Optimization of Side Impact Based on BP Network Mode

Zhou Lihui;Cheng Aiguo;Chen Tao;Zhao Min;Zhou Ze

State Key Laboratory of Advanced Design and Manufacture for Vehicle Body,Hunan University, Changsha, 410082

Online:2012-09-10 Published:2012-09-12

摘要/Abstract

摘要：

采用拉丁方试验设计方法对汽车侧围关键吸能部件的材料和板料厚度进行了多参数空间的样本数据设计,结合近似技术和多目标遗传算法,提出了一种基于BP神经网络模型的侧碰多目标优化方法。研究结果表明：该方法在满足侧碰法规GB20071的基础上,既能提高侧围结构对乘员的保护性能,又能实现其轻量化目标;BP神经网络模型的精度和效率较好地满足了工程要求。

关键词: 轻量化, 拉丁方试验设计, 近似技术, BP神经网络, 多目标遗传算法

Abstract:

The sampling data of thickness and material of automobile's side body key absorbing energy parts were obtained by the latin experimental design method. By combining approximate technology and multi-objective genetic algorithm, a multi-objective optimization method of side impact structure was established based on BP network model. The results show that: this method meets the requirements of GB20071, it improves the side structure's occupant protection capability, and achieves the lightweight objective, the accuracy and efficiency of BP network model meet the requirements of engineering design.

Key words: lightweight, latin experimental design, approximate technology, BP network, multi-objective genetic algorithm

中图分类号:

U467.1

周利辉, 成艾国, 陈涛, 赵敏, 周泽. 基于BP神经网络的侧碰多目标优化设计[J]. 中国机械工程, 2012, 23(17): 2122-2127.

ZHOU Li-Hui, CHENG Ai-Guo, CHEN Chao, DIAO Min, ZHOU Ze. #br# Multi-objective Optimization of Side Impact Based on BP Network Mode[J]. China Mechanical Engineering, 2012, 23(17): 2122-2127.

参考文献

[1]Shi Yuliang, Zhu Ping, Shen Libing, et al. Light- weight Design of Automotive Front Side Rails with TWB Concept[J]. Thin-Walled Structures, 2007, 45 : 8-14.
[2]田浩彬 [1],林建平 [1],刘瑞同 [1],许永超 [2].汽车车身轻量化及其相关成形技术综述[J].汽车工程,2005,27(3):381-384.
[3]张勇[1],李光耀[1],王建华[1].多目标遗传算法在整车轻量化优化设计中的应用研究[J].中国机械工程,2009,10(4):500-503.
[4]Myers R H, Montgomery D C. Response Surface Methodology: Process and Product Optimization Using Designed Experiments[M]. New York:John Wiley& Sons, INC. ,2002.
[5]Forsberg J, Nilsson L. Evaluation of Response Sur- face Methodologies Used in Crashworthiness Opti- mization[J]. International Journal of Impact Engi- neering, 2006, 32(5):759-777.
[6]赵敏[1],刘迪辉[1],钟志华[1],高晖[1].基于C—NCAP的类菱形概念车侧碰安全性研究[J].中国机械工程,2010(1):99-103.
[7]张维刚[1],廖兴涛[1],钟志华[1].基于逐步回归模型的汽车碰撞安全性多目标优化[J].机械工程学报,2007,43(8):142-147.
[8]邹琳[1],夏巨谌[2].混合多目标遗传算法及其在热挤压模具型腔形状优化设计中的应用[J].中国机械工程,2006,17(21):2261-2265.
[9]Deb K,Pratap A,Agarwal S,et al. A Fast and Elit- ist Multi- objective Genetic Algorithm: NSGA- Ⅱ [J]. IEEE Transactions on Evolutionary Computa- tion, 2002,6 (2) : 182-197.
[10]Zhong Z H. Finite Element Procedures for Contact - impact Problems[M]. Oxford : Oxford Universi- ty Press, 1993.
[11]Fang H,Ra Is-Rohani M,Liu Z,et al. A Compar- ative Study of Meta-modeling Methods for Multi -- objective Crashworthiness Optimization [ J ]. Computers & Structures, 2005,83 (25/26) : 2121 - 2136.
[12]Shi Q, Hagiwara I. Optimal Design Method to Au- tomobile Problems Using Holographic Neural Net- work's Approximation[J]. Japan Journal of Indus- trial and Applied Mathematics, 2000,17 (3) : 321- 339.
[13]Kitahara M , Aehenbach J D, Guo Q C. Neural Network for Crack-dapth Determination from Ul- trasonic Back-scattering Data[J]. Review of Pro- gress in Quantitative Nondestructive Evaluation, 1992, 11: 701-708.
[14]方开泰[1].均匀试验设计的理论、方法和应用——历史回顾[J].数理统计与管理,2004,23(3):69-80.
[15]丛爽,向微.BP网络结构、参数及训练方法的设计与选择[J].计算机工程,2001,27(10):36-38.
[16]王国春[1],成艾国[1],胡朝辉[1],钟志华[1].基于Kriging模型的汽车前部结构的耐撞性优化[J].汽车工程,2011(3):208-212.
[17]张君媛 [1],张敏 [1],丁如芳 [1],邱少波 [2],张雨 [2],李红建 [2].汽车侧面碰撞乘员约束系统的多目标优化[J].汽车技术,2005(11):1-4.