journal article·2026-06-26Open access

A Physics-Based Deep Learning Approach for Estimating Mechanical Properties of Layered Media Using Seismograms

Luís Pereira, Luís Godinho, Fernando G. Branco, Paulo da Venda Oliveira, Pedro Alves Costa, Aires Colaço

Applied Sciences· Volume 16· Issue 13

Abstract

This research proposes a physics-based deep learning framework, developed as a proof-of-concept based on synthetic data, for estimating the mechanical properties of layered media—namely density (ρ), Young’s modulus (E), and top layer thickness (h1)—using synthetic seismogram images generated via Finite Element Method (FEM) simulations. The dataset, comprising 5000 simulations, incorporates physical constraints and empirical density–modulus correlations. While a ResNet-style Convolutional Neural Network (CNN) extracts density and stiffness parameters from composite time–frequency images, the estimation of h1 utilizes a direct time-domain raw-signal approach to preserve spatial resolution. A 5-fold nested cross-validation scheme with internal Bayesian Optimization ensures rigorous model evaluation, further validated by normality assessments and bootstrap confidence intervals. Performance was tested against synthetic Gaussian noise (0% to 50%) and benchmarked against classical Full Waveform Inversion (FWI). The results demonstrate high predictive accuracy for shallow properties, with R2 values reaching 0.96 for Young’s modulus and 0.83 for raw-signal thickness. The neural network model requires 0.035 s per inference compared to 180 s for the FWI approach, avoiding the local minima convergence issues typical of iterative inversion. The framework exhibits resilience under moderate noise levels (up to 30%), establishing a reliable baseline for future experimental validation.
Source: crossref · Quality: baseline

Research context

OpenAlex topics: Seismic Imaging and Inversion Techniques · Seismic Waves and Analysis · Seismology and Earthquake Studies

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

Department of Civil Engineering, ISISE, ARISE, University of Coimbra, 3030-788 Coimbra, Portugal

CONSTRUCT, Faculdade de Engenharia, Universidade do Porto, 4200-465 Porto, Portugal

Funding

FCT/MCTES through national funds (PIDDAC) under the R&D Unit Institute for Sustainability and Innovation in Structural Engineering · Associate Laboratory Advanced Production and Intelligent Systems ARISE · FCT—Foundation for Science and Technology · FCT/MCTES under the R&D Unit Institute for Sustainability and Innovation in Structural Engi-neering · CONSTRUCT—Instituto de I&D em Estruturas e Construções · Fundação para a Ciência e a Tecnologia

References (46)

Show source-provided references
1. Prediction of soil properties using fuzzy membership values10.1016/j.geoderma.2010.05.001 ↗
2. Pitilakis, K.D., Anastasiadis, A., and Raptakis, D. (1992, January 19–24). Field and laboratory determination of dynamic properties of natural soil deposits. Proceedings of the 10th World Conference on Earthquake Engineering, Madrid, Spain.
3. Determination of soil properties with visible to near-and mid-infrared spectroscopy: Effects of spectral variable selection10.1016/j.geoderma.2014.01.013 ↗
4. A systematic review on inverse GRA methodologies developed for the determination of dynamic soil properties using downhole seismic array records10.1007/s40098-021-00571-2 ↗
5. Shao, W., Yue, W., Zhang, Y., Zhou, T., Zhang, Y., Dang, Y., Wang, H., Feng, X., and Chao, Z. (2023). The application of machine learning techniques in geotechnical engineering: A review and comparison. Mathematics, 11.10.3390/math11183976 ↗
6. Modeling the mechanical behavior of carbonate sands using artificial neural networks and support vector machines10.1061/(ASCE)GM.1943-5622.0000509 ↗
7. Machine learning applications in civil engineering
8. Prediction of zeolite-cement-sand unconfined compressive strength using polynomial neural network10.1140/epjp/i2016-16108-5 ↗
9. Applying machine learning and particle swarm optimization for predictive modeling and cost optimization in construction project management10.1007/s42107-023-00843-7 ↗
10. Modal Parameter Identification Method for Structural Health Monitoring Benchmark Model
11. Bilotta, A., Pollastro, A., Di Cristinzi, I., and Pecce, M.R. (2026). Comparative Study of Modal Curvature and AI-Based Approaches for Vibration-Based Damage Detection in Structural Health Monitoring Systems of Prestressed Concrete Beams. Infrastructures, 11.10.3390/infrastructures11050164 ↗
12. Prediction of secant shear modulus and damping ratio for an extremely dilative silica sand based on machine learning techniques10.1016/j.soildyn.2022.107708 ↗
13. Rock mass structural recognition from drill monitoring technology in underground mining using discontinuity index and machine learning techniques10.1016/j.ijmst.2023.02.004 ↗
14. Landslide susceptibility mapping based on self-organizing-map network and extreme learning machine10.1016/j.enggeo.2017.04.013 ↗
15. The application of deep learning in bridge health monitoring: A literature review10.1186/s43251-022-00078-7 ↗
16. Development of mathematically motivated hybrid soft computing models for improved predictions of ultimate bearing capacity of shallow foundations10.1016/j.jrmge.2022.04.005 ↗
17. Slope stability machine learning predictions on spatially variable random fields with and without factor of safety calculations10.1016/j.compgeo.2022.105094 ↗
18. Liquefaction potential assessment of soils using machine learning techniques: A state-of-the-art review from 1994–202110.1061/IJGNAI.GMENG-7788 ↗
19. Liu, H., Lin, P., and Wang, J. (2023). Machine learning approaches to estimation of the compressibility of soft soils. Front. Earth Sci., 11.10.3389/feart.2023.1147825 ↗
20. Tobler’s first law and spatial analysis10.1111/j.1467-8306.2004.09402005.x ↗
21. Machine learning for the geosciences: Challenges and opportunities10.1109/TKDE.2018.2861006 ↗
22. Machine learning in earthquake seismology10.1146/annurev-earth-071822-100323 ↗
23. Artificial neural network-based seismic detector10.1785/BSSA0850010308 ↗
24. Seismology with dark data: Image-based processing of analog records using machine learning for the Rangely earthquake control experiment10.1785/0220180298 ↗
25. Machine learning aspects of the MyShake global smartphone seismic network10.1785/0220180309 ↗
26. Intelligent real-time earthquake detection by recurrent neural networks10.1109/TGRS.2020.2966012 ↗
27. Physics-informed neural networks (PINNs) for wave propagation and full waveform inversions10.1029/2021JB023120 ↗
28. A neural network method for time-dependent inverse source problem with limited-aperture data10.1016/j.cam.2022.114842 ↗
29. An Online Interactive Physics-Informed Diffusion-Adversarial Network for Solving Mean Field Games10.1016/j.jcp.2026.114700 ↗
30. Chapra, S.C., and Canale, R.P. (2016). Métodos Numéricos Para Engenharia, McGraw Hill Brasil. [7th ed.].
31. Teixeira-Dias, F., Pinho-da-Cruz, J., Valente, R.F., and de Sousa, R.A. (2010). Método dos Elementos Finitos: Técnicas de Simulação Numérica em Engenharia, ETEP—Edições Técnicas e Profissionais.
32. Zienkiewicz, O.C., Taylor, R.L., and Zhu, J.Z. (2005). The Finite Element Method: Its Basis and Fundamentals, Elsevier.
33. Rao, S.S. (2010). The Finite Element Method in Engineering, Elsevier.
34. 3D FEM analysis of the effect of buried phononic crystal barriers on vibration mitigation10.1016/j.engstruct.2019.109340 ↗
35. Choi, Y., Ahn, D., Lee, Y., and Ahn, J. (2020). Compaction quality monitoring of open-graded aggregates by light weight deflectometer and soil stiffness gauge. Sustainability, 12.10.3390/su12062521 ↗
36. Machine learning for structural engineering: A state-of-the-art review10.1016/j.istruc.2022.02.003 ↗
37. Prediction of Blast-Induced Ground Vibration with ANN and Prediction Performance
38. Enhancing property prediction and process optimization in building materials through machine learning: A review10.1016/j.commatsci.2023.112031 ↗
39. Venkatesan, R., and Li, B. (2017). Convolutional Neural Networks in Visual Computing: A Concise Guide, CRC Press.10.4324/9781315154282 ↗
40. Goodfellow, I., Bengio, Y., Courville, A., and Bengio, Y. (2016). Deep Learning, MIT Press.
41. Nielsen, M.A. (2015). Neural Networks and Deep Learning, Determination Press.
42. Advancements and challenges in the application of artificial intelligence in civil engineering: A comprehensive review10.1007/s42107-023-00760-9 ↗
43. Pereira, L., Godinho, L., Branco, F.G., da Venda Oliveira, P., Alves Costa, P., and Colaço, A. (2025). Physics-Based Machine Learning for Vibration Mitigation by Open Buried Trenches. Appl. Sci., 15.10.3390/app152111609 ↗
44. Aki, K., and Richards, P.G. (2002). Quantitative Seismology, University Science Books. [2nd ed.].
45. Spectral-element simulations of global seismic wave propagation—I. Validation10.1046/j.1365-246X.2002.01653.x ↗
46. The finite-difference and finite-element modeling of seismic wave propagation and earthquake motion10.2478/v10155-010-0084-x ↗