PSI - Issue 64

Christoph Brenner et al. / Procedia Structural Integrity 64 (2024) 1240–1247 Christoph Brenner et al./ Structural Integrity Procedia 00 (2019) 000 – 000

1247

8

feature selection techniques and evaluating their impact on prediction accuracy is crucial. Moreover, devising methods to mitigate challenges posed by parameter correlation could lead to more precise predictions. 5. Conclusion The investigations conducted here, focusing on a case study involving a large-scale bridge with a cracked hanger connection, have demonstrated the potential enhancement of predictive maintenance for bridges through the utilization of data-driven methods. However, several challenges have emerged that necessitate attention in the future. It is crucial to devise an effective approach capable of managing a multitude of parameters, recognizing that parameters within a specific environment have similar influences on the measurements of a global SHM system. Separate predictions for individual parameters without considering their correlation concerning sensor measurements result in a notably diminished accuracy. Furthermore, real damage, especially if not yet extensive, only minimally affects stiffness and subsequently sensor readings. This aspect gains particular significance when utilizing actual measured values, which may be susceptible to distortion from measurement noise. In general, achieving sufficient accuracy with classification trees demands an extensive model database, especially for large and intricate structures like bridges, which can only be feasibly accomplished through reduced-order models. In summary, the development of an interpretable model selector based on a database of physics-based models depicting various damage states emerges as a promising avenue for advancing data-driven methods in predictive maintenance. Acknowledgements The research presented in this paper is being conducted within the project ”Digital twin as an intermediary between in- situ damage detection and global structural analysis”. The project is part of the Priority Programme SPP 2388 ”Hundred plus - Extending the Lifetime of Complex Engineering Structures through Intelligent Digitalization”, funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 501823987. References Bertsimas, D., Dunn, J., 2017. Optimal classification trees. Mach Learn 106, 1039 – 1082. Fernandez-Navamuel, A., Zamora-Sánchez, D., Omella, Á.J., Pardo, D., Garcia-Sanchez, D., Magalhães, F., 2022. Supervised Deep Learning with Finite Element simulations for damage identification in bridges. Engineering Structures 257, 114016. Huynh, D.B.P., Knezevic, D.J., Patera, A.T., 2013. A static condensation reduced basis element method: Complex problems. Computer Methods in Applied Mechanics and Engineering 259, 197 – 216. Interpretable AI LLC, 2023. Interpretable AI Documentation. Kapteyn, M.G., Knezevic, D.J., Huynh, D., Tran, M., Willcox, K.E., 2022. Data‐driven physics‐based digital twins via a library of component‐based reduced‐order models. Numerical Meth Engineering 123, 2986 – 3003. Niederer, S.A., Sacks, M.S., Girolami, M., Willcox, K., 2021. Scaling digital twins from the artisanal to the industrial. Nat Comput Sci 1, 313 – 320. Okasha, N.M., Frangopol, D.M., Orcesi, A.D., 2012. Automated finite element updating using strain data for the lifetime reliability assessment of bridges. Reliability Engineering & System Safety 99, 139 – 150. Ritto, T.G., Rochinha, F.A., 2021. Digital twin, physics-based model, and machine learning applied to damage detection in structures. Mechanical Systems and Signal Processing 155, 107614. Svendsen, B.T., Øiseth, O., Frøseth, G.T., Rønnquist, A., 2023. A hybrid structural health monitoring approach for damage detection in steel bridges under simulated environmental conditions using numerical and experimental data. Structural Health Monitoring 22, 540 – 561. Torzoni, M., Tezzele, M., Mariani, S., Manzoni, A., Willcox, K.E., 2024. A digital twin framework for civil engineering structures. Computer Methods in Applied Mechanics and Engineering 418, 116584.