Full Text:   <827>

Summary:  <784>

Suppl. Mater.: 

CLC number: TU599

On-line Access: 2021-06-21

Received: 2020-06-01

Revision Accepted: 2020-08-26

Crosschecked: 2021-05-20

Cited: 0

Clicked: 1663

Citations:  Bibtex RefMan EndNote GB/T7714

 ORCID:

Miao Su

https://orcid.org/0000-0002-8750-7204

Shao-fan Li

https://orcid.org/0000-0002-6950-1474

-   Go to

Article info.
Open peer comments

Journal of Zhejiang University SCIENCE A 2021 Vol.22 No.6 P.427-440

http://doi.org/10.1631/jzus.A2000245


Application of an interpretable artificial neural network to predict the interface strength of a near-surface mounted fiber-reinforced polymer to concrete joint


Author(s):  Miao Su, Hui Peng, Shao-fan Li

Affiliation(s):  School of Civil Engineering, Changsha University of Science and Technology, Changsha 410114, China; more

Corresponding email(s):   shaofan@berkeley.edu

Key Words:  Fiber-reinforced polymer (FRP), Bond strength, Machine learning (ML), Neural interpretation diagram (NID), Regression, Feature importance, Connection weights approach


Miao Su, Hui Peng, Shao-fan Li. Application of an interpretable artificial neural network to predict the interface strength of a near-surface mounted fiber-reinforced polymer to concrete joint[J]. Journal of Zhejiang University Science A, 2021, 22(6): 427-440.

@article{title="Application of an interpretable artificial neural network to predict the interface strength of a near-surface mounted fiber-reinforced polymer to concrete joint",
author="Miao Su, Hui Peng, Shao-fan Li",
journal="Journal of Zhejiang University Science A",
volume="22",
number="6",
pages="427-440",
year="2021",
publisher="Zhejiang University Press & Springer",
doi="10.1631/jzus.A2000245"
}

%0 Journal Article
%T Application of an interpretable artificial neural network to predict the interface strength of a near-surface mounted fiber-reinforced polymer to concrete joint
%A Miao Su
%A Hui Peng
%A Shao-fan Li
%J Journal of Zhejiang University SCIENCE A
%V 22
%N 6
%P 427-440
%@ 1673-565X
%D 2021
%I Zhejiang University Press & Springer
%DOI 10.1631/jzus.A2000245

TY - JOUR
T1 - Application of an interpretable artificial neural network to predict the interface strength of a near-surface mounted fiber-reinforced polymer to concrete joint
A1 - Miao Su
A1 - Hui Peng
A1 - Shao-fan Li
J0 - Journal of Zhejiang University Science A
VL - 22
IS - 6
SP - 427
EP - 440
%@ 1673-565X
Y1 - 2021
PB - Zhejiang University Press & Springer
ER -
DOI - 10.1631/jzus.A2000245


Abstract: 
Accurately estimating the interfacial bond capacity of the near-surface mounted (NSM) carbon fiber-reinforced polymer (CFRP) to concrete joint is a fundamental task in the strengthening and retrofit of existing reinforced concrete (RC) structures. The machine learning (ML) approach may provide an alternative to the commonly used semi-empirical or semi-analytical methods. Therefore, in this work we have developed a predictive model based on an artificial neural network (ANN) approach, i.e. using a back propagation neural network (BPNN), to map the complex data pattern obtained from an NSM CFRP to concrete joint. It involves a set of nine material and geometric input parameters and one output value. Moreover, by employing the neural interpretation diagram (NID) technique, the BPNN model becomes interpretable, as the influence of each input variable on the model can be tracked and quantified based on the connection weights of the neural network. An extensive database including 163 pull-out testing samples, collected from the authors’ research group and from published results in the literature, is used to train and verify the ANN. Our results show that the prediction given by the BPNN model agrees well with the experimental data and yields a coefficient of determination of 0.957 on the whole database. After removing one non-significant feature, the BPNN becomes even more computationally efficient and accurate. In addition, compared with the existed semi-analytical model, the ANN-based approach demonstrates a more accurate estimation. Therefore, the proposed ML method may be a promising alternative for predicting the bond strength of NSM CFRP to concrete joint for structural engineers.

应用可解释的人工神经网络预测表层嵌贴纤维增强复合材料板条与混凝土界面的粘结强度

目的:针对采用纤维增强复合材料加固的钢筋混凝土结构,本文旨在运用机器学习方法取代目前广泛使用的半经验-半分析理论公式,以准确预测该类加固结构中表层嵌贴纤维增强复合材料(CFRP)板条与混凝土界面的粘结强度.
创新点:1. 建立反向传播人工神经网络(BPNN)预测表层嵌贴CFRP板与混凝土界面的粘结强度.2. 采用基于Garson算法和连接权重算法的神经解释图(NID)定量分析神经网络中各个输入变量的重要性.
方法:1. 从作者课题组完成的实验和已发表的文献中收集共163组表层嵌贴CFRP-混凝土单剪实验结果,并形成数据集.2. 运用建立的数据集训练和测试BPNN,构建实验参数与界面粘结强度间的非线性映射关系及预测模型.3. 基于Garson算法和连接权重算法分别计算神经网络输入变量的重要性,并通过NID分析数据集中有重要影响的输入变量和无效输入变量.
结论:1. 建立的BPNN模型得出的预测结果与实验数据吻合良好,预测值与真实值之间的决定系数在整个数据集中的表现为0.957.2. 通过删除数据集中的无效输入变量可提高BPNN的计算效率和准确性.3. 与现有的半经验-半分析理论公式相比,本文建立的BPNN模型可以得出更准确的估计.

关键词:纤维增强复合材料;粘结强度;机器学习;神经解释图;回归;变量重要性;连接权重算法

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

[1]Abuodeh OR, Abdalla JA, Hawileh RA, 2020. Prediction of shear strength and behavior of RC beams strengthened with externally bonded FRP sheets using machine learning techniques. Composite Structures, 234:111698.

[2]Ali MSM, Oehlers DJ, Griffith MC, et al., 2008. Interfacial stress transfer of near surface-mounted FRP-to-concrete joints. Engineering Structures, 30(7):1861-1868.

[3]Al-Mahaidi R, Kalfat R, 2018. Rehabilitation of Concrete Structures with Fiber-reinforced Polymer. Butterworth-Heinemann, Oxford, UK.

[4]Beck MW, 2018. NeuralNetTools: visualization and analysis tools for neural networks. Journal of Statistical Software, 85(11).

[5]Bergstra J, Bengio Y, 2012. Random search for hyper-parameter optimization. Journal of Machine Learning Research, 13(10):281-305.

[6]Bilotta A, di Ludovico M, Nigro E, 2011. FRP-to-concrete interface debonding: experimental calibration of a capacity model. Composites Part B: Engineering, 42(6):1539-1553.

[7]Ceroni F, 2010. Experimental performances of RC beams strengthened with FRP materials. Construction and Building Materials, 24(9):1547-1559.

[8]Chen C, Cheng LJ, 2016. Theoretical solution to fatigue bond stress distribution of NSM FRP reinforcement in concrete. Composites Part B: Engineering, 99:453-464.

[9]Chen YF, Ding DQ, Zhu CH, et al., 2019. Size- and edge-effect cohesive energy and shear strength between graphene, carbon nanotubes and nanofibers: continuum modeling and molecular dynamics simulations. Composite Structures, 208:150-167.

[10]de Lorenzis L, Teng JG, 2007. Near-surface mounted FRP reinforcement: an emerging technique for strengthening structures. Composites Part B: Engineering, 38(2):119-143.

[11]Garson GD, 1991. Interpreting neural-network connection weights. AI Expert, 6(4):46-51.

[12]Ghasemi H, Brighenti R, Zhuang XY, et al., 2014. Optimization of fiber distribution in fiber reinforced composite by using NURBS functions. Computational Materials Science, 83:463-473.

[13]Godoy C, I. Depina I, Thakur V, 2020. Application of machine learning to the identification of quick and highly sensitive clays from cone penetration tests. Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering), 21(6):445-461. https://doi.org./10.1631/jzus.A1900556

[14]Hait P, Sil A, Choudhury S, 2020. Seismic damage assessment and prediction using artificial neural network of RC building considering irregularities. Journal of Structural Integrity and Maintenance, 5(1):51-69.

[15]Hoang ND, 2019. Estimating punching shear capacity of steel fibre reinforced concrete slabs using sequential piecewise multiple linear regression and artificial neural network. Measurement, 137:58-70.

[16]Ibrahim OM, 2013. A comparison of methods for assessing the relative importance of input variables in artificial neural networks. Journal of Applied Sciences Research, 9(11):5692-5700.

[17]Jung Y, 2018. Multiple predicting K-fold cross-validation for model selection. Journal of Nonparametric Statistics, 30(1):197-215.

[18]Karaci A, Yaprak H, Ozkaraca O, et al., 2019. Estimating the properties of ground-waste-brick mortars using DNN and ANN. CMES-Computer Modeling in Engineering & Sciences, 118(1):207-228.

[19]Mahal M, Täljsten B, Blanksvärd T, 2016. Experimental performance of RC beams strengthened with FRP materials under monotonic and fatigue loads. Construction and Building Materials, 122:126-139.

[20]Nielsen MA, 2015. Neural Networks and Deep Learning. Determination Press, New York, USA.

[21]Olden JD, Jackson DA, 2002a. A comparison of statistical approaches for modelling fish species distributions. Freshwater Biology, 47(10):1976-1995.

[22]Olden JD, Jackson DA, 2002b. Illuminating the black box: a randomization approach for understanding variable contributions in artificial neural networks. Ecological Modelling, 154(1-2):135-150.

[23]Olden JD, Joy MK, Death RG, 2004. An accurate comparison of methods for quantifying variable importance in artificial neural networks using simulated data. Ecological Modelling, 178(3-4):389-397.

[24]Özesmi SL, Özesmi U, 1999. An artificial neural network approach to spatial habitat modelling with interspecific interaction. Ecological Modelling, 116(1):15-31.

[25]Padierna LC, Carpio M, Rojas A, et al., 2017. Hyper-parameter tuning for support vector machines by estimation of distribution algorithms. In: Melin P, Castillo O, Kacprzyk J (Eds.), Nature-Inspired Design of Hybrid Intelligent Systems. Springer, Cham, Germany, p.787-800.

[26]Pedregosa F, Varoquaux G, Gramfort A, et al., 2011. Scikit-learn: machine learning in Python. The Journal of Machine Learning Research, 12:2825-2830.

[27]Peng H, Liu Y, Cai CS, et al., 2019. Experimental investigation of bond between near-surface-mounted CFRP strips and concrete under freeze-thawing cycling. Journal of Aerospace Engineering, 32(1):04018125.

[28]Petersen RB, Masia MJ, Seracino R, 2010. In-plane shear behavior of masonry panels strengthened with NSM CFRP strips. II: finite-element model. Journal of Composites for Construction, 14(6):764-774.

[29]Raschka S, Mirjalili V, 2017. Python Machine Learning: Machine Learning and Deep Learning with Python, Scikit-Learn, and TensorFlow, 2nd Edition. Packt Publishing, Birmingham, UK.

[30]Rumelhart DE, Hinton GE, Williams RJ, 1986. Learning representations by back-propagating errors. Nature, 323(6088):533-536.

[31]Sadoun O, Merdas A, Douadi A, 2020. The bond and flexural strengthening of reinforced concrete elements strengthened with near surface mounted prestressing steel (PS) bars. Journal of Adhesion Science and Technology, 34(19):2120-2143.

[32]Seracino R, Jones NM, Ali SM, et al., 2007a. Bond strength of near-surface mounted FRP strip-to-concrete joints. Journal of Composites for Construction, 11(4):401-409.

[33]Seracino R, Saifulnaz MRR, Oehlers DJ, 2007b. Generic debonding resistance of EB and NSM plate-to-concrete joints. Journal of Composites for Construction, 11(1):62-70.

[34]Shishegaran A, Khalili MR, Karami B, et al., 2020. Computational predictions for estimating the maximum deflection of reinforced concrete panels subjected to the blast load. International Journal of Impact Engineering, 139:103527.

[35]Siu C, 2017. Day32: Variable Importance in ANNs. https://csiu.github.io/blog/update/2017/03/28/day32.html

[36]Su M, Peng H, Yuan M, et al., 2021a. Identification of the interfacial cohesive law parameters of FRP strips externally bonded to concrete using machine learning techniques. Engineering Fracture Mechanics, 247:107643.

[37]Su M, Zhong Q, Peng H, et al., 2021b. Selected machine learning approaches for predicting the interfacial bond strength between FRPs and concrete. Construction and Building Materials, 270: 121456.

[38]Vu DT, Hoang ND, 2016. Punching shear capacity estimation of FRP-reinforced concrete slabs using a hybrid machine learning approach. Structure and Infrastructure Engineering, 12(9):1153-1161.

[39]Wu YF, Zhou ZQ, Yang QD, et al., 2010. On shear bond strength of FRP-concrete structures. Engineering Structures, 32(3):897-905.

[40]Yang L, Qi CC, Lin XS, et al., 2019. Prediction of dynamic increase factor for steel fibre reinforced concrete using a hybrid artificial intelligence model. Engineering Structures, 189:309-318.

[41]Zhang C, Zhao JH, Rabczuk T, 2018. The interface strength and delamination of fiber-reinforced composites using a continuum modeling approach. Composites Part B: Engineering, 137:225-234.

[42]Zhang SS, 2012. Behaviour and Modelling of RC Beams Strengthened in Flexure with Near-surface Mounted FRP Strips. PhD Thesis, The Hong Kong Polytechnic University, Hong Kong, China.

[43]Zhang SS, Teng JG, Yu T, 2013. Bond-slip model for CFRP strips near-surface mounted to concrete. Engineering Structures, 56:945-953.

[44]Zhang SS, Teng JG, Yu T, 2014. Bond strength model for CFRP strips near-surface mounted to concrete. Journal of Composites for Construction, 18(3):A4014003.

[45]Zhang SS, Yu T, Chen GM, 2017. Reinforced concrete beams strengthened in flexure with near-surface mounted (NSM) CFRP strips: current status and research needs. Composites Part B: Engineering, 131:30-42.

[46]Zhu H, Wu G, Zhang L, et al., 2014. Experimental study on the fire resistance of RC beams strengthened with near-surface-mounted high-Tg BFRP bars. Composites Part B: Engineering, 60:680-687.

Open peer comments: Debate/Discuss/Question/Opinion

<1>

Please provide your name, email address and a comment





Journal of Zhejiang University-SCIENCE, 38 Zheda Road, Hangzhou 310027, China
Tel: +86-571-87952783; E-mail: cjzhang@zju.edu.cn
Copyright © 2000 - Journal of Zhejiang University-SCIENCE