Life prediction of glass fiber reinforced plastics based on BP neural network under corrosion condition

doi:10.11949/0438-1157.20190299

Abstract

Abstract:

The factors affecting the service life of composites were analyzed by the changes of macroscopic, microscopic and mechanical properties before and after aging of glass fiber reinforced plastics under corrosive conditions. The analysis shows that the service life of fiberglass reinforced plastics under corrosion conditions is affected by three factors of temperature, time and corrosion medium concentration. Based on the bending strength retention rate of composites, a three-layer BP neural network model with a structure of 3-10-1 is used to predict the service life of composites. Through the comparison and error analysis of the forecast data and the measured data, and the random extraction of 6 sets of test data for detection, the results show that the predicted value obtained by BP Neural network model has a good fitting fit with the measured value.

Key words: fiber reinforced plastics, life prediction, neural network, corrosion, composites

摘要：

通过腐蚀条件下玻璃纤维增强塑料老化前后宏观、微观形貌及力学性能的变化对复合材料使用寿命的影响因素进行分析，分析表明，腐蚀条件下玻璃纤维增强塑料使用寿命受温度、时间和腐蚀介质浓度三种因素影响。结合玻璃纤维增强塑料的弯曲强度保留率建立结构为3-10-1的三层BP神经网络模型对复合材料使用寿命进行预测。通过预测数据和实测数据的对比及误差分析，并随机抽取6组检验数据进行检测，结果表明，所建立的BP神经网络模型得到的预测值与实测值具有较好的拟合度。

关键词: 纤维增强塑料, 寿命预测, 神经网络, 腐蚀, 复合材料

CLC Number:

TB 332

Tao WANG, Jun WANG, Diyu ZHAO, Yujian LIU, Ruigang HOU. Life prediction of glass fiber reinforced plastics based on BP neural network under corrosion condition[J]. CIESC Journal, 2019, 70(12): 4872-4880.

王涛, 王俊, 赵迪宇, 刘育建, 侯锐钢. 基于BP神经网络的玻璃纤维增强塑料腐蚀条件下的寿命预测[J]. 化工学报, 2019, 70(12): 4872-4880.

Figures/Tables 9

Fig.1 BP neural network model structure

Fig.2 Appearance of materials under different temperature corrosion conditions

Fig.3 Microscopic morphology of materials under different acid concentration corrosion conditions

Fig.4 Change curves of bending strength retention rate under different concentration of acid corrosion

Fig.5 Variation curves of flexural strength retention rate of composites with time and temperature under different acid concentration corrosion conditions

Table 1 Prediction results and measured values of neural networks under low concentration acid corrosion

No.	Time/d	Temperature/℃	Acid concentration $/ (m o l · L - 1)$	Predictive	Actual/%	Normalization	Relative error/%
1	0	25	2	1.0852	100.00	1.0000	8.5200
2	1	25	2	0.9979	98.05	0.9701	2.8633
3	3	25	2	1.0185	95.76	0.9352	8.9111
4	7	25	2	0.8758	90.60	0.8563	2.2706
5	14	25	2	0.8830	89.80	0.8441	4.6071
6	21	25	2	0.9103	90.31	0.8520	6.8467
7	90	25	2	0.8744	86.33	0.7911	10.5340
8	180	25	2	0.6320	85.73	0.7820	-19.1824
9	360	25	2	0.7403	85.30	0.7753	-4.5143
10	0	55	2	0.9764	100.00	1.0000	-2.3610
11	1	55	2	0.9353	97.86	0.9673	-3.3083
12	3	55	2	0.9257	95.32	0.9285	-0.2992
13	7	55	2	0.8226	88.94	0.8309	-1.0029
14	14	55	2	0.8050	85.97	0.7856	2.4751
15	21	55	2	0.7986	84.28	0.7597	5.1188
16	60	55	2	0.6584	73.83	0.6001	9.7206
17	90	55	2	0.7251	71.10	0.5584	29.8466
18	180	55	2	0.5093	68.44	0.5177	-1.6097
19	360	55	2	0.3993	61.96	0.4187	-4.6440
20	0	80	2	0.9611	100.00	1.0000	-3.8889
21	1	80	2	0.5951	84.41	0.7618	-21.879
22	3	80	2	0.6647	78.24	0.6674	-0.4162
23	7	80	2	0.5673	72.23	0.5756	-1.4477
24	14	80	2	0.4948	68.21	0.5142	-3.7814
25	21	80	2	0.4491	66.49	0.4880	-7.9697
26	30	80	2	0.5171	64.58	0.4588	12.7147
27	60	80	2	0.4637	59.45	0.3804	8.5200
28	90	80	2	0.3371	55.53	0.3205	2.8633
29	180	80	2	0.2072	47.87	0.2033	8.9111
30	360	80	2	0.0386	40.91	0.0970	2.2706

Table 1 Prediction results and measured values of neural networks under low concentration acid corrosion

No.	Time/d	Temperature/℃	Acid concentration $/ (m o l · L - 1)$	Predictive	Actual/%	Normalization	Relative error/%
1	0	25	2	1.0852	100.00	1.0000	8.5200
2	1	25	2	0.9979	98.05	0.9701	2.8633
3	3	25	2	1.0185	95.76	0.9352	8.9111
4	7	25	2	0.8758	90.60	0.8563	2.2706
5	14	25	2	0.8830	89.80	0.8441	4.6071
6	21	25	2	0.9103	90.31	0.8520	6.8467
7	90	25	2	0.8744	86.33	0.7911	10.5340
8	180	25	2	0.6320	85.73	0.7820	-19.1824
9	360	25	2	0.7403	85.30	0.7753	-4.5143
10	0	55	2	0.9764	100.00	1.0000	-2.3610
11	1	55	2	0.9353	97.86	0.9673	-3.3083
12	3	55	2	0.9257	95.32	0.9285	-0.2992
13	7	55	2	0.8226	88.94	0.8309	-1.0029
14	14	55	2	0.8050	85.97	0.7856	2.4751
15	21	55	2	0.7986	84.28	0.7597	5.1188
16	60	55	2	0.6584	73.83	0.6001	9.7206
17	90	55	2	0.7251	71.10	0.5584	29.8466
18	180	55	2	0.5093	68.44	0.5177	-1.6097
19	360	55	2	0.3993	61.96	0.4187	-4.6440
20	0	80	2	0.9611	100.00	1.0000	-3.8889
21	1	80	2	0.5951	84.41	0.7618	-21.879
22	3	80	2	0.6647	78.24	0.6674	-0.4162
23	7	80	2	0.5673	72.23	0.5756	-1.4477
24	14	80	2	0.4948	68.21	0.5142	-3.7814
25	21	80	2	0.4491	66.49	0.4880	-7.9697
26	30	80	2	0.5171	64.58	0.4588	12.7147
27	60	80	2	0.4637	59.45	0.3804	8.5200
28	90	80	2	0.3371	55.53	0.3205	2.8633
29	180	80	2	0.2072	47.87	0.2033	8.9111
30	360	80	2	0.0386	40.91	0.0970	2.2706

Table 2 Prediction results and measured values of neural networks under the condition of high concentration acid corrosion

No.	Time/d	Temperature/℃	Acid concentration/( $m o l · L - 1$ )	Predictive	Actual/%	Normalization	Relative error/%
1	0	25	7	1.0952	100.00	1.0000	9.5200
2	1	25	7	0.9116	92.40	0.8839	3.1428
3	3	25	7	0.9183	89.20	0.8350	9.9804
4	7	25	7	0.8498	88.90	0.8304	2.3416
5	14	25	7	0.8030	84.56	0.7641	5.0897
6	21	25	7	0.7970	82.90	0.7387	7.8969
7	30	25	7	0.7632	79.05	0.6799	12.2574
8	90	25	7	0.6997	70.53	0.5497	27.2883
9	180	25	7	0.4409	65.70	0.4759	-7.3552
10	360	25	7	0.3636	59.90	0.3872	-6.0972
11	0	55	7	0.9680	100.00	1.0000	-3.2000
12	1	55	7	0.8495	90.33	0.8523	-0.3259
13	3	55	7	0.8342	89.70	0.8426	-0.9890
14	7	55	7	0.6828	80.51	0.7022	-2.7690
15	14	55	7	0.5506	73.14	0.5895	-6.5970
16	21	55	7	0.5210	72.47	0.5793	-10.070
17	60	55	7	0.2904	64.47	0.4571	-36.461
18	90	55	7	0.4428	64.08	0.4511	-1.8474
19	180	55	7	0.3361	55.28	0.3167	6.1404
20	360	55	7	0.2794	50.30	0.2405	16.1678
21	0	80	7	1.1667	100.00	1.0000	16.6667
22	1	80	7	0.7461	83.20	0.7433	0.3737
23	3	80	7	0.7081	80.35	0.6997	1.1909
24	7	80	7	0.6623	76.63	0.6429	3.0246
25	14	80	7	0.4315	65.34	0.4704	-8.2681
26	21	80	7	0.4051	64.89	0.4635	-12.586
27	30	80	7	0.3395	62.23	0.4228	-19.708
28	60	80	7	0.3006	55.32	0.3172	-5.2538
29	180	80	7	0.0759	39.78	0.0798	-4.8751
30	360	80	7	-0.0583	34.56	0	—

Table 2 Prediction results and measured values of neural networks under the condition of high concentration acid corrosion

No.	Time/d	Temperature/℃	Acid concentration/( $m o l · L - 1$ )	Predictive	Actual/%	Normalization	Relative error/%
1	0	25	7	1.0952	100.00	1.0000	9.5200
2	1	25	7	0.9116	92.40	0.8839	3.1428
3	3	25	7	0.9183	89.20	0.8350	9.9804
4	7	25	7	0.8498	88.90	0.8304	2.3416
5	14	25	7	0.8030	84.56	0.7641	5.0897
6	21	25	7	0.7970	82.90	0.7387	7.8969
7	30	25	7	0.7632	79.05	0.6799	12.2574
8	90	25	7	0.6997	70.53	0.5497	27.2883
9	180	25	7	0.4409	65.70	0.4759	-7.3552
10	360	25	7	0.3636	59.90	0.3872	-6.0972
11	0	55	7	0.9680	100.00	1.0000	-3.2000
12	1	55	7	0.8495	90.33	0.8523	-0.3259
13	3	55	7	0.8342	89.70	0.8426	-0.9890
14	7	55	7	0.6828	80.51	0.7022	-2.7690
15	14	55	7	0.5506	73.14	0.5895	-6.5970
16	21	55	7	0.5210	72.47	0.5793	-10.070
17	60	55	7	0.2904	64.47	0.4571	-36.461
18	90	55	7	0.4428	64.08	0.4511	-1.8474
19	180	55	7	0.3361	55.28	0.3167	6.1404
20	360	55	7	0.2794	50.30	0.2405	16.1678
21	0	80	7	1.1667	100.00	1.0000	16.6667
22	1	80	7	0.7461	83.20	0.7433	0.3737
23	3	80	7	0.7081	80.35	0.6997	1.1909
24	7	80	7	0.6623	76.63	0.6429	3.0246
25	14	80	7	0.4315	65.34	0.4704	-8.2681
26	21	80	7	0.4051	64.89	0.4635	-12.586
27	30	80	7	0.3395	62.23	0.4228	-19.708
28	60	80	7	0.3006	55.32	0.3172	-5.2538
29	180	80	7	0.0759	39.78	0.0798	-4.8751
30	360	80	7	-0.0583	34.56	0	—

Table 3 Prediction results and measured values of neural networks for test data

Test	Time/d	Temperature/℃	Acid concentration/( $m o l · L - 1$ )	Predictive/%	Actual/%	Relative error/%
1	30	25	2	92.32	90.26	2.28
2	30	55	2	75.39	80.33	-6.15
3	90	80	2	60.32	55.53	8.63
4	60	25	7	73.23	77.43	-5.42
5	30	55	7	70.31	70.97	-0.93
6	90	80	7	38.36	45.54	-15.77

Table 3 Prediction results and measured values of neural networks for test data

Test	Time/d	Temperature/℃	Acid concentration/( $m o l · L - 1$ )	Predictive/%	Actual/%	Relative error/%
1	30	25	2	92.32	90.26	2.28
2	30	55	2	75.39	80.33	-6.15
3	90	80	2	60.32	55.53	8.63
4	60	25	7	73.23	77.43	-5.42
5	30	55	7	70.31	70.97	-0.93
6	90	80	7	38.36	45.54	-15.77

Fig.6 Comparison of predictive and actual of test data

References 30

1	秦川丽, 张巨生, 唐冬雁, 等 . 聚氨酯/乙烯基酯树脂互穿聚合物网络合成动力学及相容性[J]. 化工学报, 2004, 55(2): 253-258.
	Qin C L , Zhang J S , Tang D Y , et al . Kinetics and compatibility of polyurethane/vinyl ester interpenetrating polymer networks[J]. Journal of Chemical Industry and Engineering(China), 2004, 55(2): 253-258.
2	Sen R , Mullins G . Application of FRP composites for underwater piles repair[J]. Composites Part B Engneering, 2007, 38(5): 751-758.
3	Monti M , Natali M , Petrucci R , et al . Carbon nanofibers for strain and impact damage sensing in glass fiber reinforced composites based on an unsaturated polyester resin[J]. Polymer Composites, 2011, 32(5): 766-775.
4	Alia C , Jofre-Reche J A , Suárez J C , et al . Characterization of the chemical structure of vinyl ester resin in a climate chamber under different conditions of degradation[J]. Polymer Degradation & Stability, 2018, 153: 88-99.
5	Yadav S , Gangwar S , Singh S . Micro/nano reinforced filled metal alloy composites: a review over current development in aerospace and automobile applications[J]. Materials Today Proceedings, 2017, 4(4): 5571-5582.
6	徐国强, 代礼葵, 孙耀宁 . 碱性腐蚀对玻纤增强树脂基复合材料性能的影响[J]. 工程塑料应用, 2018， (11): 94-99.
	Xu G Q , Dai L K , Sun Y N . Effects of alkaline corrosion on properties of glass fiber reinforced resin matrix composites[J]. Engineering Plastics Applications, 2018， (11): 94-99.
7	Malkapuram R , Kumar V , Negi Y S . Recent development in natural fiber reinforced polypropylene composites[J]. Journal of Reinforced Plastics & Composites, 2009, 28(10): 1169-1189.
8	Hollaway L C . Handbook of Polymer Composites for Engineers[M]. Elsevier Science, 1994.
9	程基伟, 王天民 . TWINTEX纤维增强塑料的应力腐蚀开裂[J]. 材料研究学报, 2005, (3): 269-276.
	Cheng J W , Wang T M . Stress corrosion cracking of TWINTEX fiber reinforced plastics [J]. Journal of Materials Research, 2005, (3): 269-276.
10	何伟, 廖功雄, 靳奇峰, 等 . 新型特种工程塑料聚芳醚砜酮和聚芳醚砜热分解动力学及寿命预测[J]. 化工学报, 2006, 57(4): 981-986.
	He W , Liao G X , Jin Q F , et al . Thermal decomposition kinetics and life prediction of poly(aryl ether sulfone ketone) and poly(aryl ether sulfone) new engineering plastics [J].Journal of Chemical Industry and Engineering(China), 2006, 57(4): 981-986.
11	Xu P , Yates S J , Nino J C . Hydrothermal corrosion of magnesia-pyrochlore composites for inert matrix materials[J]. Journal of Composite Materials, 2010, 44(12): 1533-1545.
12	Husnayain F , Latif M , Garniwa I . Transformer oil lifetime prediction using the Arrhenius law based on physical and electrical characteristics[C]// 2015 International Conference on Quality in Research (QiR). Lombok, Indonesia, 2016.
13	Kuo H C , Wu L J . Prediction of heat-affected zone using grey theory[J]. Journal of Materials Processing Tech., 2002, 120(1): 151-168.
14	Guzman V A , Brøndsted P . Effects of moisture on glass fiber-reinforced polymer composites[J]. Journal of Composite Materials. 2015, 49(8): 911-920.
15	Yu Y H , Li P , Sui G , et al . Effects of hygrothermal aging on the thermal-mechanical properties of vinylester resin and its pultruded carbon fiber composites[J]. Polymer Composites， 2009, 30(10): 1458-1464.
16	韩露, 马芳武, 陈实现, 等 . 玄武岩纤维增强聚乳酸力学性能及耐老化性能[J]. 化工学报, 2019, 70(3): 1171-1178.
	Han L , Ma F W , Chen S X , et al . Mechanical properties and aging resistance of basalt fiber reinforced polylactic acid [J]. CIESC Journal, 2019, 70(3): 1171-1178.
17	Hsu K L , Gupta H V , Sorooshian S . Artificial neural network modeling of the rainfall-runoff process[J]. Water Resources Research, 1995, 31(10): 2517-2530.
18	Park D C , El-Sharkawi M A , Marks R J , et al . Electric load forecasting using an artificial neural network[J]. IEEE Transactions on Power Systems, 2013, 6(2): 442-449.
19	Vassilopoulos A P , Georgopoulos E F , Dionysopoulos V . Artificial neural networks in spectrum fatigue life prediction of composite materials[J]. International Journal of Fatigue, 2007, 29(1): 20-29.
20	郑晶晶, 张钦礼, 王新民, 等 . 基于BP神经网络的充填钻孔使用寿命预测[J]. 矿业工程研究, 2008, 30(4): 40-44.
	Zheng J J , Zhang Q L , Wang X M , et al . Life prediction of filling boreholes based on BP neural network [J]. Mining Engineering Research, 2008, 30 (4): 40-44.
21	胡中永, 邬友英, 李艳华 . GB/T 1449—2005：纤维增强塑料弯曲性能试验方法[S]. 北京: 中国标准出版社. 2005.
	Hu Z Y , Hu Y Y , Li Y H . GB/T 1449—2005： Test Method for Bending Properties of Fiber Reinforced Plastics [S]. Beijing: China Standard Press， 2005.
22	Zhu M X , Zhang D L . Study on the algorithms of selecting the radial basis function center[J]. Journal of Anhui University, 2000, 24(1): 78.
23	蔡斌, 许龙飞, 郝丽妍 . 基于BP神经网络的受火后RC梁钢筋温度预测[J]. 北方建筑, 2019, 4(4): 23-27.
	Cai B , Xu L F , Hao L Y . Prediction of RC beam rebar temperature after fire based on BP neural network [J]. Northern Architecture, 2019, 4 (4): 23-27.
24	张斌 . 金属板料单点渐进成形的回弹研究[D].西安: 陕西科技大学, 2019.
	Zhang B . Research on springback of sheet metal single point incremental forming [D]. Xi an: Shaanxi University of Science and Technology, 2019.
25	于桂杰, 赵崇, 迟建伟, 等 . 基于人工神经网络的连续油管疲劳寿命预测[J]. 中国石油大学学报(自然科学版), 2018, 227(3): 136-141.
	Yu G J , Zhao C , Chi J W , et al . Fatigue life prediction of coiled tubing based on artificial neural network [J]. Journal of China University of Petroleum (Natural Science Edition), 2018, 227(3): 136-141.
26	李晓骏, 许凤和, 陈新文 . 先进聚合物基玻璃纤维增强塑料的热氧老化研究[J]. 材料工程, 1999, (12): 19-22+30.
	Li X J , Xu F H , Chen X W . Study on thermo-oxygen aging of advanced polymer matrix composites [J]. Material Engineering, 1999, (12): 19-22+30.
27	Inácio A L N , Nonato R C , Bonse B C . Mechanical and thermal behavior of aged composites of recycled PP/EPDM/talc reinforced with bamboo fiber[J]. Polymer Testing, 2018， DOI: 10.1016/j.polymertesting.2018.10.035 DOI
28	栗晓飞, 张琦, 项民, 等 . 炭纤维增强树脂基玻璃纤维增强塑料模拟加速腐蚀方法的研究[J]. 材料工程, 2016, (2): 5-9.
	Li X F , Zhang Q , Xiang M , et al . Study on the simulated accelerated corrosion method of carbon fiber reinforced resin matrix composites [J]. Material Engineering, 2016, (2): 5-9.
29	Putić S , Stamenović M , Petrović J , et al . Effect of alkaline solutions on the tensile properties of glass-polyester pipes[J]. Materials & Design, 2011, 32(4): 2456-2461.
30	侯锐钢, 尚琪冬, 黎大胜 . 混酸介质加速老化条件下玻璃纤维/溴化环氧乙烯基酯复合材料的耐久性[J]. 复合材料学报, 2017, 34(6): 1212-1220.
	Hou R G , Shang Q D , Li D S . Durability of glass fiber/brominated vinyl epoxy ester composites under accelerated ageing in mixed acid media [J]. Journal of Composite Materials, 2017, 34(6): 1212-1220.

[1]	Fei KANG, Weiguang LYU, Feng JU, Zhi SUN. Research on discharge path and evaluation of spent lithium-ion batteries [J]. CIESC Journal, 2023, 74(9): 3903-3911.
[2]	Kaijie WEN, Li GUO, Zhaojie XIA, Jianhua CHEN. A rapid simulation method of gas-solid flow by coupling CFD and deep learning [J]. CIESC Journal, 2023, 74(9): 3775-3785.
[3]	Gang YIN, Yihui LI, Fei HE, Wenqi CAO, Min WANG, Feiya YAN, Yu XIANG, Jian LU, Bin LUO, Runting LU. Early warning method of aluminum reduction cell leakage accident based on KPCA and SVM [J]. CIESC Journal, 2023, 74(8): 3419-3428.
[4]	Xingzhi HU, Haoyan ZHANG, Jingkun ZHUANG, Yuqing FAN, Kaiyin ZHANG, Jun XIANG. Preparation and microwave absorption properties of carbon nanofibers embedded with ultra-small CeO₂ nanoparticles [J]. CIESC Journal, 2023, 74(8): 3584-3596.
[5]	Jiaqi CHEN, Wanyu ZHAO, Ruichong YAO, Daolin HOU, Sheying DONG. Synthesis of pistachio shell-based carbon dots and their corrosion inhibition behavior on Q235 carbon steel [J]. CIESC Journal, 2023, 74(8): 3446-3456.
[6]	Chengying ZHU, Zhenlei WANG. Operation optimization of ethylene cracking furnace based on improved deep reinforcement learning algorithm [J]. CIESC Journal, 2023, 74(8): 3429-3437.
[7]	Linqi YAN, Zhenlei WANG. Multi-step predictive soft sensor modeling based on STA-BiLSTM-LightGBM combined model [J]. CIESC Journal, 2023, 74(8): 3407-3418.
[8]	Ye XU, Wenjun HUANG, Junpeng MI, Chuanchuan SHEN, Jianxiang JIN. Surge diagnosis method of centrifugal compressor based on multi-source data fusion [J]. CIESC Journal, 2023, 74(7): 2979-2987.
[9]	Ao ZHANG, Yingwu LUO. Low modulus, high elasticity and high peel adhesion acrylate pressure sensitive adhesives [J]. CIESC Journal, 2023, 74(7): 3079-3092.
[10]	Jing ZHAO, Chengwen GU, Xigao JIAN, Zhihuan WENG. Preparation and performance evaluation of magnolol-based epoxy resin anti-corrosion coatings [J]. CIESC Journal, 2023, 74(7): 3010-3017.
[11]	Bin CAI, Xiaolin ZHANG, Qian LUO, Jiangtao DANG, Liyuan ZUO, Xinmei LIU. Research progress of conductive thin film materials [J]. CIESC Journal, 2023, 74(6): 2308-2321.
[12]	Yanhui LI, Shaoming DING, Zhouyang BAI, Yinan ZHANG, Zhihong YU, Limei XING, Pengfei GAO, Yongzhen WANG. Corrosion micro-nano scale kinetics model development and application in non-conventional supercritical boilers [J]. CIESC Journal, 2023, 74(6): 2436-2446.
[13]	Xuejin GAO, Yuzhuo YAO, Huayun HAN, Yongsheng QI. Fault monitoring of fermentation process based on attention dynamic convolutional autoencoder [J]. CIESC Journal, 2023, 74(6): 2503-2521.
[14]	Lei HUANG, Lingxue KONG, Jin BAI, Huaizhu LI, Zhenxing GUO, Zongqing BAI, Ping LI, Wen LI. Effect of oil shale addition on ash fusion behavior of Zhundong high-sodium coal [J]. CIESC Journal, 2023, 74(5): 2123-2135.
[15]	Jialin DAI, Weidong BI, Yumei YONG, Wenqiang CHEN, Hanyang MO, Bing SUN, Chao YANG. Effect of thermophysical properties on the heat transfer characteristics of solid-liquid phase change for composite PCMs [J]. CIESC Journal, 2023, 74(5): 1914-1927.