BP神经网络预测复合材料热压罐成型均匀性

林源; 关志东

doi:10.13700/j.bh.1001-5965.2020.0158

BP神经网络预测复合材料热压罐成型均匀性

doi: 10.13700/j.bh.1001-5965.2020.0158

林源,
关志东^,

北京航空航天大学航空科学与工程学院, 北京 100083

详细信息

通讯作者:
关志东, E-mail: zdguan@buaa.edu.cn

中图分类号: V258⁺.3;TB332
计量
- 文章访问数: 479
- HTML全文浏览量: 70
- PDF下载量: 58
- 被引次数: 0
出版历程
- 收稿日期: 2020-04-23
- 录用日期: 2020-05-30
- 网络出版日期: 2021-06-20

Predicting the formation uniformity of composite autoclave by BP neural network

LIN Yuan,
GUAN Zhidong^,

School of Aeronautic Science and Engineering, Beihang University, Beijing 100083, China

More Information

Corresponding author: GUAN Zhidong, E-mail: zdguan@buaa.edu.cn

摘要

摘要:
复合材料热压罐成型过程中的固化度差值是复合材料固化度均匀性的主要表征参数之一。基于3层BP神经网络，以复合材料双平台固化工艺曲线的加热速率、保温时间和保温温度为输入参数，建立了成型过程任一时刻最大固化度差值的快速估算模型。仿真复合材料热压罐成型过程，得到最大固化度差值作为试验样本数据，对BP神经网络进行训练，训练结束后对该模型的准确性进行验证。结果表明：该BP神经网络估算模型准确性和效率较高，为复合材料热压罐成型最大固化度差值的估算提供了一种快速有效的新方法。
- 复合材料 /
- 神经网络 /
- 估算 /
- 固化工艺 /
- 热压罐 /
- 残余应力
Abstract:
The difference in degree of cure in the forming process of composite autoclave is one of the main characterization parameters of degree of cure uniformity of composite. Based on thethree-layer BP neural network, this paper established a rapid estimation model of maximum difference of curing degree at any time in the forming process with heating rate, holding time and holding temperature as input parameters. Maximum difference in degree of cure was obtained by simulating the forming process of composite autoclave as test sample data to train the BP neural network, and the accuracy of the model was verified after the training. The results show that the accuracy and efficiency of this BP neural network model are high, which provides a fast and effective new method for estimating the difference of the maximum curing degree of composite autoclave.
- composite /
- neural network /
- estimation /
- curing process /
- autoclave /
- residual stress

HTML全文

图 1 BP神经网络模型结构

Figure 1. Structure of BP neural network model

下载: 全尺寸图片幻灯片

图 2 BP神经网络估算模型训练均方误差曲线

Figure 2. Training MSE curves of BP neural network estimation model

下载: 全尺寸图片幻灯片

表 1 复合材料AS4/3501-6热力学及固化动力学模型参数

Table 1. Thermodynamics and cure kinetics model parameters for AS4/3501-6 composites

参数	数值
ρ/(kg·m^-3)	1 578
C_p/(J·(kg·K)^-1)	862
k_T/(W·(m·K)^-1)	0.413 5
k_L/(W·(m·K)^-1)	12.83
H_r/(J·kg^-1)	198.6×10³
A₁/min^-1	2.102×10⁹
A₂/min^-1	-2.014×10⁹
A₃/min^-1	1.960×10⁵
ΔE₁/(J·mol^-1)	8.07×10⁴
ΔE₂/(J·mol^-1)	7.78×10⁴
ΔE₃/(J·mol^-1)	5.66×10⁴
注：ρ为复合材料的密度; C_p为比热; k_T和k_L分别为复合材料的横向和纵向导热系数; H_r为最终反应热; A_i(i=1, 2, 3)为频率因子; ΔE_i(i=1, 2, 3)为活化能。

下载: 导出CSV

表 2 工艺参数取值范围

Table 2. Range of process parameters

参数	取值范围
a₁/(℃·min^-1)	[1,5]
a₂/(℃·min^-1)	[1,5]
T₁/℃	[115,155]
T₂/℃	[175,215]
t_p/min	[0, 100]
t_g/min	[0, 150]

下载: 导出CSV

表 3 固化数据

Table 3. Curing data

a₁/(℃·min^-1)	a₂/(℃·min^-1)	T₁/℃	T₂/℃	t_p/min	t_g/min	Δα
3	3	120	180	120	60	0.008 23
3	3	121	180	120	60	0.008 23
3	3	119	180	120	60	0.008 36
4	3	120	179	120	60	0.008 04
2	3	120	179	120	60	0.007 96
2	4	120	179	120	60	0.008 50
2	2	120	179	120	60	0.006 78
2	2	120	179	123	60	0.006 78
1	1	116	176	102	72	0.003 48
1	1	116	175	93	78	0.003 42
1	1	116	176	93	78	0.003 42
1	2	116	175	93	78	0.005 89
1	1	116	175	90	78	0.003 42
1	1	116	175	96	78	0.003 41
1	1	116	175	96	80	0.003 39
1	1	115	177	66	100	0.003 26
1	1	121	175	78	100	0.002 92
1	1	121	177	78	100	0.002 92
1	1	130	175	78	100	0.002 33
1	1	130	176	78	100	0.002 34

下载: 导出CSV

表 4 Δα的估算

Table 4. Estimation of Δα

a₁/(℃·min^-1)	a₂/(℃·min^-1)	T₁/℃	T₂/℃	t_p/min	t_g/min	Δα的试验值Δα_s	Δα的估算值Δα_g	误差值Δ=Δα_s-Δα_g	误差率Δ/Δ_s
3	3	120	181	120	60	0.008 56	0.008 09	0.000 47	0.054
2	2	120	179	123	60	0.006 78	0.006 53	0.000 25	0.036
1	1	119	178	120	60	0.003 82	0.003 89	-0.000 07	-0.018
1	1	117	117	114	64	0.003 69	0.003 66	0.000 03	0.008
1	1	115	175	90	88	0.003 38	0.003 50	-0.000 12	-0.035

下载: 导出CSV

参考文献(22)

[1]	DOLKUN D, ZHU W D, XU Q, et al. Optimization of cure profile for thick composite parts based on finite element analysis and genetic algorithm[J]. Journal of Composite Materials, 2018, 52(28): 1-5. http://www.researchgate.net/publication/324697272_Optimization_of_cure_profile_for_thick_composite_parts_based_on_finite_element_analysis_and_genetic_algorithm
[2]	顾轶卓, 李敏, 李艳霞, 等. 飞行器结构用复合材料制造技术与工艺理论进展[J]. 航空学报, 2015, 36(8): 2773-2797. https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201508025.htm GU Y Z, LI M, LI Y X, et al. Progress on manufacturing technology and process theory of aircraft composite[J]. Acta Aeronautica et Astronautica Sinica, 2015, 36(8): 2773-2797(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201508025.htm
[3]	王永贵, 梁宪珠, 曹正华, 等. 热压罐工艺成型先进复合材料构件的温度场研究综述[J]. 玻璃钢/复合材料, 2009(3): 81-85. doi: 10.3969/j.issn.1003-0999.2009.03.021 WANG Y G, LIANG X Z, CAO Z H, et al. Review of the temperature field research of autoclave moulding for advanced composite components[J]. Fiber Reinforced Plastics/Composites, 2009(3): 81-85(in Chinese). doi: 10.3969/j.issn.1003-0999.2009.03.021
[4]	李彩林. 复合材料热压罐固化设计的数值模拟[J]. 玻璃钢/复合材料, 2014(11): 26-29. https://www.cnki.com.cn/Article/CJFDTOTAL-BLGF201411005.htm LI C L. Numerical simulation for autoclave curing design of composite materials[J]. Fiber Reinforced Plastics/Composites, 2014(11): 26-29(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-BLGF201411005.htm
[5]	向炳东, 李敏, 李艳霞, 等. 筒状复合材料制件热压罐成型温度模拟及影响因素分析[J]. 航空学报, 2017, 38(11): 298-307. https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201711028.htm XIANG B D, LI M, LI Y X, et al. Numerical simulation and parameter analysis of temperature distribution of autoclave cured composite cylindrical structure[J]. Acta Aeronautica et Astronautica Sinica, 2017, 38(11): 298-307(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201711028.htm
[6]	LOOS A C, SPRINGER G S. Curing of epoxy matrix composites[J]. Journal of Composite Materials, 1983, 17(2): 135-169. doi: 10.1177/002199838301700204
[7]	BOGETTI T A, GILLESPIE J W. Two-dimensional cure simulation of thick thermosetting composites[J]. Journal of Composite Materials, 1991, 25(3): 239-273. doi: 10.1177/002199839102500302
[8]	JOHNSTON A A. An integrated model of the development of process-induced deformation in autoclave processing of composite structures[D]. Ann Arbor: The University of British Columbia (Canada), 1998.
[9]	HOA S V. Design and manufacturing of composites[C]//Proceedings of the Second Joint Hamada Concordia University Montreal, 1998, 16(1): 2-16.
[10]	焦李成. 神经网络系统理论[M]. 西安: 西安电子科技大学出版社, 1990. JIAO L C. Theory of neural network system[M]. Xi'an: Xidian University Press, 1990(in Chinese).
[11]	吴简彤, 王建华. 神经网络技术及其应用[M]. 哈尔滨: 哈尔滨工程大学出版社, 1998. WU J T, WANG J H. Neural network technology and its application[M]. Harbin: Harbin Engineering University Press, 1998(in Chinese).
[12]	张德丰, 丁伟雄, 雷晓平. MATLAB程序设计与综合应用[M]. 北京: 清华大学出版社, 2012: 1-3. ZHANG D F, DING W X, LEI X P. MATLAB programming and comprehensive application[M]. Beijing: Tsinghua University Press, 2012: 1-3(in Chinese).
[13]	BLEST D C, DUFFY B R, MCKEE S, et al. Curing simulation of thermoset composites[J]. Composites Part A: Applied Science and Manufacturing, 1999, 30(11): 1289-1309. doi: 10.1016/S1359-835X(99)00032-9
[14]	CHEUNG A, YU Y, POCHIRAJU K. Three-dimensional finite element simulation of curing of polymer composites[J]. Finite Elements in Analysis and Design, 2004, 40(8): 895-912. doi: 10.1016/S0168-874X(03)00119-7
[15]	KIM Y K, WHITE S R. Viscoelastic analysis of processing-induced residual stress in thick composite laminates[J]. Mechanics of Composite Materials and Structures, 1997, 4(4): 361-387. doi: 10.1080/10759419708945889
[16]	高腾龙. 热固性树脂基复合材料固化变形模拟及控制方法研究[D]. 西安: 西北工业大学, 2018. GAO T L. Simulation and control methods of curing deformation of thermosetting resin matrix composites[D]. Xi'an: Northwestern Polytechnical University, 2018(in Chinese).
[17]	林仕高, 欧元贤. BP神经网络学习参数优化研究[J]. 微计算信息, 2010, 26(6): 199-200. https://www.cnki.com.cn/Article/CJFDTOTAL-WJSJ201016080.htm LIN S G, OU Y X. Research of the optimization of the learning parameters in BP neural network[J]. Microcomputer Information, 2010, 26(6): 199-200(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-WJSJ201016080.htm
[18]	HECHT-NIELSEN R. Theory of the backpropagation neural network[J]. Neural Networks, 1988, 1(S1): 445. http://doi.ieeecomputersociety.org/resolve?ref_id=doi:10.1109/IJCNN.1989.118638&rfr_id=trans/ts/1992/07/tts1992070551.htm
[19]	蒋佳欣, 王博, 王猛, 等. 基于BIM技术和BP神经网络的成都理工大学图书馆天然采光研究[J]. 土木建筑工程信息技术, 2020, 12(1): 30-38. https://www.cnki.com.cn/Article/CJFDTOTAL-TMJZ202001006.htm JIANG J X, WANG B, WANG M, et al. Study on natural lighting design for CDUT library based on BIM and BP neural network[J]. Journal of Information Technology in Civil Engineering and Architecture, 2020, 12(1): 30-38(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-TMJZ202001006.htm
[20]	薛掌安, 邓海亮. BP神经网络在复合材料研究中的应用[J]. 材料导报, 2009, 23(14): 250-253. https://www.cnki.com.cn/Article/CJFDTOTAL-CLDB2009S1079.htm XUE Z A, DENG H L. Application of BP neural networks in the composites research[J]. Materials Reports, 2009, 23(14): 250-253(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-CLDB2009S1079.htm
[21]	柳小桐. BP神经网络输入层数据归一化研究[J]. 机械工程与自动化, 2010(3): 122-123. https://www.cnki.com.cn/Article/CJFDTOTAL-SXJX201003047.htm LIU X T. Study on data normalization in BP neural network[J]. Mechanical Engineering & Automation, 2010(3): 122-123(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-SXJX201003047.htm
[22]	HECHT-NIELSEN R. Kolmogorov's mapping neural network existence theorem[C]//Proceedings of the International Conference on Neural Networks, 1987, 3: 11-13.