基于神经网络的车辆抗冲击防护组件优化

李明星; 王显会; 周云波; 孙晓旺; 曾斌; 胡文海

doi:10.11883/bzycj-2019-0055

基于神经网络的车辆抗冲击防护组件优化

doi: 10.11883/bzycj-2019-0055

1.
南京理工大学机械工程学院，江苏南京 210094
2.
东风汽车公司技术中心，湖北武汉 430056

基金项目: 国家自然科学基金（11802140，51405232）；中央高校基本科研业务费专项资金（30918011303）；道路交通安全公安部重点实验室开放基金（2018ZDSYSKFKT09）

详细信息

作者简介:
李明星（1994－　），男，硕士研究生，1217280185@qq.com

通讯作者:
王显会（1968－　），男，博士，教授，13770669850@139.com

中图分类号: O385; E952; U462.2
计量
- 文章访问数: 5849
- HTML全文浏览量: 1779
- PDF下载量: 134
- 被引次数: 0
出版历程
- 收稿日期: 2019-02-27
- 修回日期: 2019-06-20
- 网络出版日期: 2019-12-25
- 刊出日期: 2020-02-01

Research on optimization of vehicle anti-shock protection components based on neural network

1.
School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, Jiangsu, China
2.
Technical Center of Dongfeng Motor Corporation, Wuhan 430056, Hubei, China

摘要

摘要: 随着军用车辆防护要求的不断提高，抗冲击防护组件设计面临越来越多的挑战。为了提供一种高效科学的研究方法，本文中采用V型结构，并应用径向基函数神经网络近似模型和多目标遗传算法对某型车防护组件进行优化设计。以防护组件变形量与总体质量最小为设计目标，利用灵敏度分析筛选出对防护组件防护性能影响较大的设计因子。以径向基函数神经网络构建实验设计样本的近似模型，然后用多目标遗传算法进行数值优化获得防护组件最优方案。最后通过仿真与实验验证，证明优化方案满足设计要求。研究结果可为今后防护组件开发提供设计思路。
- 防护组件 /
- V型结构 /
- 灵敏度分析 /
- 神经网络 /
- 多目标遗传算法
Abstract: With the increasing requirements for the protection of military vehicles, the design of impact protection components is facing more and more challenges. In order to provide an efficient and scientific research method, this paper adopts a V-shaped structure, and uses radial basis function neural network approximation model and multi-objective genetic algorithm to optimize the design of a certain type of vehicle protection components. Taking the deformation amount of the protection component and the total mass as the design goal, the sensitivity analysis is used to select the design factor that has a great influence on the protection performance of the protection component. The approximate model of the experimental design sample is constructed by radial basis function neural network, and then multi-objective genetic algorithm is used to numerically optimize the optimal component of the protection component. Finally, through simulation and experimental verification, it is proved that the optimization scheme meets the design requirements. Provide a design idea for the future development of protective components.
- protective component /
- V-shaped structure /
- sensitivity analysis /
- neural network /
- multi-objective genetic algorithm

HTML全文

图 1 防护组件结构图

Figure 1. Structure of protective component

下载: 全尺寸图片幻灯片

图 2 台架仿真有限元模型

Figure 2. Bench simulation finite element model

下载: 全尺寸图片幻灯片

图 3 数值仿真与实验中的抗爆组件压溃情况

Figure 3. The crushing situation of anti-explosion components in numerical simulation and experiment

下载: 全尺寸图片幻灯片

图 4 设计变量位置示意图

Figure 4. Design variable position diagram

下载: 全尺寸图片幻灯片

图 5 变形域图

Figure 5. Deformation domain diagram

下载: 全尺寸图片幻灯片

图 6 设计变量贡献率

Figure 6. Design variable contribution rate

下载: 全尺寸图片幻灯片

图 7 神经网络模型与Kriging模型对比

Figure 7. Comparison between neural network model and Kriging model

下载: 全尺寸图片幻灯片

图 8 主影响因子X₂ 分别与X₁、X₄、X₆和X₇拟合的位移响应面

Figure 8. Response surfaces of displacements where the main influence factor X₂ fits X₁, X₄, X₆ and X₇, respectively

下载: 全尺寸图片幻灯片

图 9 帕累托最优解集

Figure 9. Pareto optimal solution set

下载: 全尺寸图片幻灯片

图 10 抗爆组件优化后实验状态

Figure 10. Experimental status of anti-explosion components after optimization

下载: 全尺寸图片幻灯片

图 11 仿真与实验背板塑性变形图

Figure 11. Simulation and experimental plastic deformation of the back plate

下载: 全尺寸图片幻灯片

图 12 背板位移曲线

Figure 12. Backplane displacement curves

下载: 全尺寸图片幻灯片

表 1 TNT炸药的JWL方程参数

Table 1. JWL equation parameters of TNT explosives

A/GPa	B/GPa	R₁	R₂	ω	E₀/（kJ·kg⁻¹）
371	3.33	4.15	0.95	0.30	7 000

下载: 导出CSV

表 2 设计变量初始值

Table 2. Design variable initial value

变量	参数定义	初始值
X₁	面板厚度	6.00 mm
X₂	背板厚度	6.00 mm
X₃	中间横梁厚度	6.00 mm
X₄	两侧横梁厚度	5.75 mm
X₅	纵梁厚度	5.00 mm
X₆	中间前侧横梁位置	0
X₇	中间后侧横梁位置	0

下载: 导出CSV

表 3 拉丁超立方实验采样样本

Table 3. Latin hypercube test sample

No.	X₁/mm	X₂/mm	X₃/mm	X₄/mm	X₅/mm	X₆	X₇
1	8.49	5.69	4.45	5.06	4.36	−0.95	0.38
$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $
20	9.65	6.42	6.76	7.90	6.05	−0.56	−0.78
21	7.03	4.11	5.84	4.03	6.16	0.54	−0.94
$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $
39	9.33	4.09	5.79	6.49	4.86	0.16	0.12
40	9.80	7.71	5.13	6.53	4.12	−0.46	−0.18

下载: 导出CSV

表 4 优化前后对比

Table 4. Comparison before and after optimization

优化前后	模拟地板变形量/ mm	设计质量/ kg
优化前	210	303
优化后	134	328

下载: 导出CSV

参考文献(19)

[1]	王显会, 佘磊, 郭启涛, 等. 基于抗冲击波响应的新型蜂窝夹层结构多目标优化设计 [J]. 车辆与动力技术, 2014(4): 25–30. WANG X H, YAN L, GUO Q T, et al. Multi-objective optimization design of new honeycomb sandwich structure based on shock wave response [J]. Vehicle & Power Technology, 2014(4): 25–30.
[2]	张钱城, 郝方楠, 李裕春, 等. 爆炸冲击载荷作用下车辆和人员的损伤与防护 [J]. 力学与实践, 2014, 36(5): 527–539. DOI: 10.6052/1000-0879-13-539. ZHANG Q C, HAO F N, LI Y C, et al. Damage and protection of vehicles and personnel under blast loading [J]. Mechanics in Engineering, 2014, 36(5): 527–539. DOI: 10.6052/1000-0879-13-539.
[3]	KENDALE A, AMERICAS T, JATEGAONKAR R, et al. Study of occupant responses in a mine blast using MADYMO [C] // SAFE Symposium, 2009.
[4]	张中英, 何洋扬, 王乐阳, 等. 车底结构对爆炸冲击波响应特性影响研究[C] // 全国仿真技术学术会议. 九江, 2009. 123−126.
[5]	孙京帅. 蜂窝材料面内冲击吸能性能优化及在电动汽车耐撞性设计中的应用[D]. 大连: 大连理工大学, 2013.
[6]	FENG H M. Self-generation RBFNs using evolutional PSO learning [J]. Neurocomputing, 2006, 70(1-3): 241–251. DOI: 10.1016/j.neucom.2006.03.007.
[7]	柳建容, 马咏梅, 黄巍. 基BP神经网络与遗传算法的减振器优化设计 [J]. 机械科学与技术, 2011(8): 1267–1271. LIU J R, MA Y M, HUANG W. Optimal design of shock absorber based on bp neural network and genetic algorithm [J]. Mechanical Science and Technology, 2011(8): 1267–1271.
[8]	李利莎, 谢清粮, 郑全平, 等. 基于Lagrange、ALE和SPH算法的接触爆炸模拟计算 [J]. 爆破, 2011, 28(1): 18–22. DOI: 10.3963/j.issn.1001-487X.2011.01.005. LI L S, XIE Q L, ZHENG Q P, et al. Simulation of contact explosion based on Lagrange, ALE and SPH algorithms [J]. Blasting, 2011, 28(1): 18–22. DOI: 10.3963/j.issn.1001-487X.2011.01.005.
[9]	方开泰. 均匀实验设计的理论、方法和应用——历史回顾 [J]. 数理统计与管理, 2004, 23(3): 69–80. DOI: 10.3969/j.issn.1002-1566.2004.03.016. FANG K T. Theory, method and application of uniform experimental design—historical review [J]. Journal of Mathematical Statistics and Management, 2004, 23(3): 69–80. DOI: 10.3969/j.issn.1002-1566.2004.03.016.
[10]	SOBOL I M. Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates [M]. Elsevier Science Publishers, 2001. DOI: 10.1016/S0378-4754(00)00270-6.
[11]	CHEN S, WANG X X, BROWN D J. Sparse incremental regression modeling using correlation criterion with boosting search [J]. IEEE Signal Processing Letters, 2005, 12(3): 198–201. DOI: 10.1109/LSP.2004.842250.
[12]	CHEN S, WOLFGANG A, HARRIS C J, et al. Symmetric RBF classifier for nonlinear detection in multiple-antenna-aided systems [J]. IEEE Transactions on Neural Networks, 2008, 19(5): 737. DOI: 10.1109/TNN.2007.911745.
[13]	GONZALEZ J, ROJAS I, ORTEGA J, et al. Multiobjective evolutionary optimization of the size, shape, and position parameters of radial basis function networks for function approximation [J]. IEEE Transactions on Neural Networks, 2003, 14(6): 1478–1495. DOI: 10.1109/TNN.2003.820657.
[14]	LEUNG F H F, LAM H K, LING S H, et al. Tuning of the structure and parameters of a neural network using an improved genetic algorithm [J]. IEEE Transactions on Neural Networks, 2003, 14(1): 79–88. DOI: 10.1109/TNN.2002.804317.
[15]	BORS A G, PITAS I. Median radial basis function neural network [J]. IEEE Transactions on Neural Networks, 1996, 7(6): 1351–1364. DOI: 10.1109/72.548164.
[16]	YIN H, ALLINSON N M. Self-organizing mixture networks for probability density estimation [J]. IEEE Transactions on Neural Networks, 2001, 12(2): 405–411. DOI: 10.1109/72.914534.
[17]	RAJEEV S, KRISHNAMOORTHY C S. Genetic algorithm-based methodology for design optimization of reinforced concrete frames [J]. Computer-Aided Civil and Infrastructure Engineering, 2002, 13(1): 63–74. DOI: 10.1111/0885-9507.00086.
[18]	魏然, 王显会, 周云波, 等. 帕累托最优在车辆底部防护结构设计中的应用研究 [J]. 兵工学报, 2015, 36(6). DOI: 10.3969/j.issn.1000-1093.2015.06.013. WEI R, WANG X H, ZHOU Y B, et al. Research on the application of Pareto optimality in the protection structure design of vehicle bottom [J]. Acta Armamentarii, 2015, 36(6). DOI: 10.3969/j.issn.1000-1093.2015.06.013.
[19]	蔡亚. 多目标遗传算法的改进及其变速箱参数优化设计研究[D]. 合肥: 合肥工业大学, 2014. DOI: CNKI:CDMD:2.1015.568685.