Abstract
Data used for training structural health monitoring (SHM) systems are often expensive and/or impractical to obtain, particularly labelled data. Population-based SHM presents a potential solution to this issue by considering the available data across a population of structures. However, differences between structures will mean that the training and testing distributions will differ; thus, conventional machine learning methods cannot be expected to generalise between structures. To address this issue, transfer learning—in the form of domain adaptation (DA)—can be used to align the underlying distributions. An important consideration when applying DA is that it may lead to performance degradation; this scenario is referred to as negative transfer. Furthermore, validating whether negative transfer has occurred is challenging without labelled data, so assessing the similarity between the datasets is of critical importance to mitigate the likelihood of negative transfer. Typical unsupervised metrics measure the discrepancy between the marginal distributions, so they are not robust to scenarios where the conditional distributions differ significantly. This paper presents a discussion on negative transfer, proposing a physics-based metric that utilises the modal assurance criteria (MAC), between the modes of the healthy structures. This metric is compared to other popular metrics, showing that the proposed MAC-based metric indicates the likelihood of negative transfer under conditional distribution shift, whereas the previous unsupervised methods cannot.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
This method assumes that some normal condition data can be labelled by assuming data from the start of a monitoring campaign correspond to the healthy structure.
- 2.
This approach could be used to predict any performance metric by normalising values to \([0,1]\).
References
Dervilis, N., Choi, M., Taylor, S., Barthorpe, R., Park, G., Farrar, C., Worden, K.: On damage diagnosis for a wind turbine blade using pattern recognition. J. Sound Vib. 333(6), 1833–1850 (2014)
Farrar, C., Worden, K.: Structural Health Monitoring: A Machine Learning Perspective. Wiley, Chichester (2013)
Bull, L.A., Gardner, P., Gosliga, J., Rogers, T.J., Dervilis, N., Cross, E.J., Papatheou, E., Maguire, A.E., Campos, C., Worden, K.: Foundations of population-based SHM, Part I: Homogeneous populations and forms. Mech. Syst. Signal Process. 148, 107141 (2021)
Gosliga, J., Gardner, P., Bull, L.A., Dervilis, N., Worden, K.: Foundations of population-based SHM, Part II: heterogeneous populations – graphs, networks, and communities. Mech. Syst. Signal Process. 148, 107144 (2021)
Gardner, P., Bull, L., Gosliga, J., Dervilis, N., Worden, K.: Foundations of population-based SHM, Part III: Heterogeneous populations – Map** and transfer. Mech. Syst. Signal Process. 149, 107142 (2021)
Zhuang, F., Qi, Z., Duan, K., **, D., Zhu, Y., Zhu, H., **ong, H., He, Q.: A comprehensive survey on transfer learning. Proc. IEEE 109(1), 43–76 (2021)
Xu, S., Noh, H.Y.: Phymdan: Physics-informed knowledge transfer between buildings for seismic damage diagnosis through adversarial learning. Mech. Syst. Signal Process. 151, 107374 (2021)
Gardner, P., Bull, L., Gosliga, J., Poole, J., Dervilis, N., Worden, K.: A population-based shm methodology for heterogeneous structures: Transferring damage localisation knowledge between different aircraft wings. Mech. Syst. Signal Process. 172, 108918 (2022)
Bull, L., Gardner, P., Dervilis, N., Papatheou, E., Haywood-Alexander, M., Mills, R., Worden, K.: On the transfer of damage detectors between structures: An experimental case study. J. Sound Vib. 501, 116072 (2021)
Gardner, P., Bull, L., Dervilis, N., Worden, K.: Overcoming the problem of repair in structural health monitoring: Metric-informed transfer learning. J. Sound Vib., 116245 (2021)
Gardner, P., Liu, X., Worden, K.: On the application of domain adaptation in structural health monitoring. Mech. Syst. Signal Process. 138, 106550 (2020)
Zhang, W., Deng, L., Wu, D.: Overcoming negative transfer: A survey. ar**v, pp. 1–15 (2020)
Ben-David, S., Blitzer, J., Crammer, K., Pereira, F.: Analysis of representations for domain adaptation. Adv. Neural Inf. Process. Syst., 137–144 (2007)
Gretton, A., Sriperumbudur, B., Sejdinovic, D., Strathmann, H., Balakrishnan, S., Pontil, M., Fukumizu, K.: Optimal kernel choice for large-scale two-sample tests. Adv. Neural Inf. Process. Syst. 2, 1205–1213 (2012)
Allemang, R.J.: The modal assurance criterion–twenty years of use and abuse. Sound Vib. 37(8), 14–23 (2003)
Long, M., Wang, J., Ding, G., Pan, S.J., Yu, P.S.: Adaptation regularization: A general framework for transfer learning. IEEE Trans. Knowl. Data Eng. 26(5), 1076–1089 (2014)
Ben-David, S., Blitzer, J., Crammer, K., Kulesza, A., Pereira, F., Vaughan, J. W.: A theory of learning from different domains. Mach. Learn. 79(1-2), 151–175 (2010)
Murphy, K.P.: Machine Learning: A Probabilistic Perspective. MIT Press, Cambridge (2012)
Jialin Pan, S., Tsang, I.W., Kwok, J.T., Yang, Q.: Domain adaptation via transfer component analysis. IEEE Trans. Neural Netw. 22(2), 199–210 (2011)
Long, M., Cao, Y., Wang, J., Jordan, M.I.: Learning transferable features with deep adaptation networks. In: 32nd International Conference on Machine Learning, ICML 2015, vol. 1, pp. 97–105 (2015)
Wang, J., Chen, Y., Hao, S., Feng, W., Shen, Z.: Balanced distribution adaptation for transfer learning. In: Proceedings - IEEE International Conference on Data Mining, ICDM, pp. 1129–1134 (2017)
Ganin, Y., Lempitsky, V.: Unsupervised domain adaptation by backpropagation. In: International Conference on Machine Learning, pp. 1180–1189 (2015)
Long, M., Cao, Z., Wang, J., Jordan, M.I.: Conditional adversarial domain adaptation. Adv. Neural Inf. Process. Syst. 2018-Decem(NeurIPS), 1640–1650 (2018)
Worden, K., Dulieu-Barton, J.: An overview of intelligent fault detection in systems and structures. Struct. Health Monit. 3(1), 85–98 (2004)
Farrar, C.R., Cornwell, P.J., Doebling, S.W., Prime, M.B.: Structural health monitoring studies of the alamosa canyon and i-40 bridges. Technical report, Los Alamos National Lab.(LANL), Los Alamos, NM (United States), 2000
Dervilis, N., Worden, K., Cross, E.: On robust regression analysis as a means of exploring environmental and operational conditions for shm data. J Sound Vib. 347, 279–296 (2015)
Worden, K., Burrows, A.: Optimal sensor placement for fault detection. Eng. Struct. 23(8), 885–901 (2001)
Christides, S., Barr, A.D.S.: One-dimensional theory of cracked Bernoulli-Euler beams. Int. J. Mech. Sci. 26(11-12), 639–648 (1984)
Gong, B., Shi, Y., Sha, F., Grauman, K.: Geodesic flow kernel for unsupervised domain adaptation. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 2066–2073 (2012)
Poole, J., Gardner, P., Dervilis, N., Bull, L., Worden, K.: On statistic alignment for domain adaptation in structural health monitoring. Preprint (2022). ar**v:2205.12052
Pan, S.J., Ni, X., Sun, J.T., Yang, Q., Chen, Z.: Cross-domain sentiment classification via spectral feature alignment. In: Proceedings of the 19th International Conference on World Wide Web, WWW ’10, pp. 751–760 (2010)
Mclean, J., Jones, M., O’Connell, B., Maguire, A., Rogers, T.: Physically meaningful uncertainty quantification in probabilistic wind turbine power curve models as a damage sensitive feature. Preprint (2022). ar**v:2209.15579
Acknowledgements
The authors would like to acknowledge the support of the UK Engineering and Physical Sciences Research Council via grants EP/R006768/1, EP/R003645/1, EP/R004900/1 and EP/W005816/1. For the purpose of open access, the authors have applied for a Creative Commons Attribution (CC-BY-ND) licence to any author accepted manuscript version arising.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Society for Experimental Mechanics, Inc
About this paper
Cite this paper
Poole, J., Gardner, P., Dervilis, N., Mclean, J.H., Rogers, T.J., Worden, K. (2023). Towards Physics-Based Metrics for Transfer Learning in Dynamics. In: Madarshahian, R., Hemez, F. (eds) Data Science in Engineering, Volume 10. SEM 2023. Conference Proceedings of the Society for Experimental Mechanics Series. Springer, Cham. https://doi.org/10.1007/978-3-031-34946-1_9
Download citation
DOI: https://doi.org/10.1007/978-3-031-34946-1_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-34945-4
Online ISBN: 978-3-031-34946-1
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)