Machine learning methods used in Structural Health Monitoring applications still have generalization difficulties among structures, even when structures are nominally and topologically similar. The data sets present divergences between their probability distributions that do not allow the model’s generalization for damage detection. This issue is even more complex in situations where one wants to quantify damage levels through data sets collected from different structures. Transfer learning methods offer a solution to overcome those limitations, using relevant information from a labeled structure (source domain) to assist the analysis of another structure (target domain) under unknown conditions. Therefore, this paper proposes the use of transfer component analysis to mitigate divergences between the model/structure’s features, and the label consistency requirement is applied in combination with a Gaussian process regression model for damage quantification. The effectiveness of the estimated model improves when the labels consistency between domains is achieved, indicating the current damage level in the structure when the regression model achieves its best performance (lowest error). The proposed methodology is applied on the benchmark data of a three-story building structure from the Los Alamos National Laboratory using the knowledge from its numerical model under several conditions, where the complete information of its behavior is available. The results compare the analysis in the original space and after applying the proposed methodology, demonstrating an improvement of the performance in the damage detection and quantification steps.
YaoYDorettoG. Boosting for transfer learning with multiple sources. In: 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 13–18 June 2010, San Francisco, CA2010: 1855–1862. DOI: 10.1109/CVPR.2010.5539857.
FigueiredoEBrownjohnJ. Three decades of statistical pattern recognition paradigm for shm of bridges. Struct Health Monit2022: 147592172210752. DOI: 10.1177/14759217221075241.
8.
ZhuangFQiZDuanK, et al.A comprehensive survey on transfer learning. Proc IEEE2021; 109(1): 43–76. DOI: 10.1109/JPROC.2020.3004555.
9.
WeissKKhoshgoftaarTMWangD. A survey of transfer learning. J Big Data2016; 3(1): 9. DOI: 10.1186/s40537-016-0043-6.
10.
QuattoniACollinsMDarrellT. Transfer learning for image classification with sparse prototype representations. In: 2008 IEEE Conference on Computer Vision and Pattern Recognition, 23–25 June 2008, Anchorage, AK, pp. 1–8. DOI: 10.1109/CVPR.2008.4587637.
11.
GaoJLingHHuW, et al.Transfer learning based visual tracking with gaussian processes regression. In Computer Vision – ECCV2014. ISBN 978-3-319-10578-9, pp. 188–203.
12.
CireşanDCMeierUSchmidhuberJ. Transfer learning for latin and chinese characters with deep neural networks. In: The 2012 International Joint Conference on Neural Networks (IJCNN), pp. 1–6.
13.
ZophBYuretDMayJ, et al.Transfer Learning for Low-Resource Neural Machine Translation. arXiv preprint arXiv:160402201 2016.
14.
van OpbroekA.IkramM. A.VernooijM. W., et al. Transfer learning improves supervised image segmentation across imaging protocols. IEEE Trans Med Imaging2015; 34(5): 1018–1030. DOI: 10.1109/TMI.2014.2366792.
15.
HuynhB. Q.LiH.GigerM. L.Digital mammographic tumor classification using transfer learning from deep convolutional neural networks. J Med Imaging2016; 3(3): 034501–034505. DOI: 10.1117/1.JMI.3.3.034501.
16.
ChoiKFazekasGSandlerM, et al.Transfer Learning for Music Classification and Regression Tasks. arXiv preprint arXiv:170309179 2017.
GosligaJGardnerPABullLA, et al.Foundations of population-based shm, part II: heterogeneous populations - graphs, networks, and communities. Mech Syst Signal Process2021; 148: 107144. DOI: 10.1016/j.ymssp.2020.107144, http://www.sciencedirect.com/science/article/pii/S0888327020305306.
19.
LongMCaoZWangJ, et al.Conditional adversarial domain adaptation. arXiv preprint arXiv:170510667 2017.
20.
WangJXieJZhangL, et al.A factor analysis based transfer learning method for gearbox diagnosis under various operating conditions. In: 2016 International Symposium on Flexible Automation (ISFA), 1–3 August 2016, Cleveland, OH. IEEE, pp. 81–86
21.
LuWLiangBChengY, et al.Deep model based domain adaptation for fault diagnosis. IEEE Trans Ind Elect2017; 64(3): 2296–2305. DOI: 10.1109/TIE.2016.2627020.
NamJPanSJKimS. Transfer defect learning. In: 2013 35th International Conference on Software Engineering (ICSE), 18–26 May 2013, San Francisco, CA2013: 382–391. DOI: 10.1109/ICSE.2013.6606584.
24.
SilvaSdYanoMOGonsalez-BuenoCG. Transfer component analysis for compensation of temperature effects on the impedance-based structural health monitoring. J Nondestructive Eval2021; 40. DOI: 10.1007/s10921-021-00794-6.
TeimouriHMilaniASLoeppkyJ, et al.A gaussian process-based approach to cope with uncertainty in structural health monitoring. Struct Health Monit2017; 16(2): 174–184. DOI: 10.1177/1475921716669722.
29.
PavlackBPaixãoJda SilvaS, et al.Polynomial chaos-kriging metamodel for quantification of the debonding area in large wind turbine blades. Struct Health Monit2021; 21: 666–682. DOI: 10.1177/14759217211007956.
30.
PaixãoJda SilvaSFigueiredoE, et al.Delamination area quantification in composite structures using gaussian process regression and auto-regressive models. J Vibration Control2020; 27: 2778–2792. DOI: 10.1177/1077546320966183.
31.
FigueiredoEParkGFigueirasJ, et al.Structural Health Monitoring Algorithm Comparisons Using Standard Data Sets. Los Alamos, NM: Los Alamos National Lab.(LANL), 2009. DOI: 10.2172/961604. Technical report.
32.
PanSJYangQ. A survey on transfer learning. IEEE Trans Knowledge Data Eng2010; 22(10): 1345–1359. DOI: 10.1109/TKDE.2009.191.
33.
CookDFeuzKDKrishnanNC. Transfer learning for activity recognition: a survey. Knowledge Information Systems2013; 36(3): 537–556, DOI: 10.1007/s10115-013-0665-3.
34.
LongMZhuHWangJet al.Unsupervised domain adaptation with residual transfer networks. arXiv preprint arXiv:160204433 2016.
35.
GlorotXBordesABengioY. Domain adaptation for large-scale sentiment classification: A deep learning approach. In: Proceedings of the 28th international conference on machine learning (ICML-11), Bellevue, WA, 28 June–02 July 2011. pp. 513–520.
36.
GopalanRRuonan LiRChellappaR. Domain adaptation for object recognition: an unsupervised approach. In: 2011 International Conference on Computer Vision, 06–13 November 2011, Barcelona, Spain2011: 999–1006. DOI: 10.1109/ICCV.2011.6126344.
37.
GaninYLempitskyV. Unsupervised domain adaptation by backpropagation. arXiv preprint arXiv:14097495 2014.
38.
GhifaryMKleijnWBZhangM, et al.Deep reconstruction-classification networks for unsupervised domain adaptation. In: Computer Vision – ECCV2016, pp. 597–613. ISBN 978-3-319-46493-0.
39.
PatriciaNCaputoB. Learning to learn, from transfer learning to domain adaptation: A unifying perspective. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 18–20 June 1996, San Francisco, CA. pp. 1442–1449.
40.
PanSJTsangIWKwokJT, et al.Domain adaptation via transfer component analysis. IEEE Trans Neural Networks2011; 22(2): 199–210. DOI: 10.1109/TNN.2010.2091281.
41.
SchölkopfBSmolaAMüllerKR. Kernel principal component analysis. In Artificial Neural Networks — ICANN’97, pp. 583–588. ISBN 978-3-540-69620-9.
42.
GrubingerTBirlutiuASchönerH, et al.Domain generalization based on transfer component analysis. In: RojasIJoyaGCatalaA (eds). Advances in Computational Intelligence. Cham, Switzerland: Springer International Publishing, pp. 325–334.
43.
YuYSzepesváriC. Analysis of kernel mean matching under covariate shift. arXiv preprint arXiv:12064650 2012.
44.
HarmoucheJDelphaCDialloD. Incipient fault detection and diagnosis based on kullback-leibler divergence using principal component analysis: Part i. Signal Process2014; 94: 278–287, DOI: 10.1016/j.sigpro.2013.05.018.
45.
GrettonABorgwardtKRaschM, et al.A kernel method for the two-sample-problem. Adv Neural Information Processing Systems2006; 19: 513–520.
46.
SchölkopfB. The kernel trick for distances. In: Advances in neural information processing systems, pp. 301–307.
47.
XieJZhangLDuanL, et al.On cross-domain feature fusion in gearbox fault diagnosis under various operating conditions based on transfer component analysis. In: 2016 IEEE International Conference on Prognostics and Health Management (ICPHM), 20–22 June 2016, Ottawa, ON; 2016: 1–6. DOI: 10.1109/ICPHM.2016.7542845
48.
KangFLiuXLiJ. Temperature effect modeling in structural health monitoring of concrete dams using kernel extreme learning machines. Struct Health Monit2020; 19(4): 987–1002. DOI: 10.1177/1475921719872939.
49.
FigueiredoEParkGFarinholtKM, et al.Use of time-series predictive models for piezoelectric active-sensing in structural health monitoring applications. J Vibration Acoustics2012; 134(4).
GarlaVTaylorCBrandtC. Semi-supervised clinical text classification with laplacian svms: an application to cancer case management. J Biomed Inform2013; 46(5): 869–875. DOI: 10.1016/j.jbi.2013.06.014.
52.
LiptonZ. C.ElkanC.NaryanaswamyB., 2014. Optimal thresholding of classifiers to maximize f1 measure. In Machine Learning and Knowledge Discovery in Databases. ISBN 978-3-662-44851-9, pp. 225–239. DOI:10.1007/978-3-662-44851-9_15.
53.
RanganathanAYangM-HHoJ. Online sparse gaussian process regression and its applications. IEEE Trans Image Process2011; 20(2): 391–404. DOI: 10.1109/TIP.2010.2066984.
54.
HeHSiuW-C. Single image super-resolution using gaussian process regression. Cvpr 20112011: 449–456. DOI: 10.1109/CVPR.2011.5995713.
WilsonAAdamsR. .Gaussian process kernels for pattern discovery and extrapolation. In: Proceedings of the 30th International Conference on Machine Learning, Proceedings of Machine Learning Research, 17–19 June 2013, Atlanta, GA, 28. PMLR, pp. 1067–1075, https://proceedings.mlr.press/v28/wilson13.html.
59.
ChalupkaKWilliamsCKMurrayI. A framework for evaluating approximation methods for gaussian process regression. J Machine Learn Res2013; 14: 333–350.
60.
KangFLiJ. Displacement model for concrete dam safety monitoring via gaussian process regression considering extreme air temperature. J Struct Eng2020; 146(1): 05019001. DOI: 10.1061/(ASCE)ST.1943-541X.0002467.
61.
BandaraRPChanTHThambiratnamDP. Structural damage detection method using frequency response functions. Struct Health Monit2014; 13(4): 418–429. DOI: 10.1177/1475921714522847.
62.
NguyenTChanTHThambiratnamDP. Controlled monte carlo data generation for statistical damage identification employing mahalanobis squared distance. Struct Health Monit2014; 13(4): 461–472. DOI: 10.1177/1475921714521270.
63.
ChampneysMDGreenAMoralesJ, et al.On the vulnerability of data-driven structural health monitoring models to adversarial attack. Struct Health Monit2020: 20(4). DOI: 10.1177/1475921720920233.
64.
YanoMOVillaniLGGda SilvaS, et al.Autoregressive model extrapolation using cubic splines for damage progression analysis. J Braz Soc Mech Sci Eng2021; 43(1): 1–15. DOI: 10.1007/s40430-020-02734-3.
65.
EntezamiAShariatmadarHKaramodinA. Data-driven damage diagnosis under environmental and operational variability by novel statistical pattern recognition methods. Struct Health Monit2019; 18(5–6): 1416–1443. DOI: 10.1177/1475921718800306.
66.
JaynesET. Information theory and statistical mechanics. Phys Rev1957; 106: 620–630. DOI: 10.1103/PhysRev.106.620.
67.
SunHBettiR. A hybrid optimization algorithm with bayesian inference for probabilistic model updating. Computer-Aided Civil Infrastructure Eng2015; 30(8): 602–619. DOI: 10.1111/mice.12142.
