pith. sign in

arxiv: 2606.27304 · v1 · pith:7DZGFJ2Knew · submitted 2026-06-25 · 💻 cs.LG

A Multi-Fidelity Convolutional Autoencoder-Transfer Learning Framework for Guided-Wave-Based Damage Diagnosis Using Large Simulated and Limited Experimental Datasets

Santosh Kapuria , Abhishek This is my paper

Pith reviewed 2026-06-26 05:05 UTC · model grok-4.3

classification 💻 cs.LG
keywords guided wave structural health monitoringconvolutional autoencodertransfer learningmulti-fidelitydamage localizationspectral element modelpiezoelectric transducers
0
0 comments X

The pith

Convolutional autoencoder transfer learning from large synthetic guided-wave datasets to limited experimental measurements achieves R² scores above 0.93 for damage localization.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a multi-fidelity framework that pretrains a convolutional autoencoder on large datasets generated by a one-dimensional time-domain spectral element model of guided waves, then uses transfer learning to adapt the model to a small number of labeled experimental measurements for damage localization and sizing in plate structures. This addresses the practical barriers of scarce experimental data and expensive high-fidelity simulations in guided-wave structural health monitoring. A sympathetic reader would care because the approach is shown to outperform a CNN baseline while generalizing to previously unseen damage scenarios. The reported performance metrics indicate that such models could become viable for onboard monitoring systems where only limited real-world data can be collected.

Core claim

The CAE-based transfer learning framework significantly outperforms its CNN-based counterpart in damage localisation accuracy. The model achieves excellent predictive performance with R² scores exceeding 0.93 for damage localisation and 0.99 for damage sizing. Its generalisation capability is demonstrated on previously unseen data, showing high prediction accuracy for damage scenarios not represented during pretraining or fine-tuning. The results establish the proposed framework as an accurate, computationally efficient, and practically viable solution for real-world GWSHM applications.

What carries the argument

Convolutional autoencoder pretrained on synthetic data from a one-dimensional time-domain spectral element model, followed by transfer learning with a feed-forward network and limited experimental data to diagnose damage.

If this is right

  • The framework supplies accurate damage localization and sizing even when only a small quantity of experimental data is available.
  • It outperforms CNN-based methods specifically in localization accuracy.
  • The trained model maintains high prediction accuracy on damage scenarios absent from both the synthetic pretraining and the experimental fine-tuning sets.
  • The overall approach reduces computational cost relative to full high-fidelity simulation while remaining suitable for practical onboard monitoring.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the synthetic-to-experimental feature transfer proves robust, the same pretraining strategy could be tested on other wave-based sensing problems that face similar data scarcity.
  • Systematic variation of the number of fine-tuning samples would reveal the minimum experimental data volume required for acceptable performance.
  • Extending the underlying simulation model beyond one dimension could address potential limits when structures have more complex geometries.

Load-bearing premise

Features learned from the one-dimensional spectral element simulations transfer effectively to real experimental guided-wave signals when the network is fine-tuned on a small set of labeled measurements.

What would settle it

If R² scores on a held-out experimental test set with new damage locations and sizes fall below 0.8 or if the CAE framework fails to exceed the CNN baseline on the same data, the transfer-learning claim would be disproven.

Figures

Figures reproduced from arXiv: 2606.27304 by Abhishek, Santosh Kapuria.

Figure 1
Figure 1. Figure 1: High-level flow chart for training and testing phases of the proposed CAE-FFNN-TL model for guided wave-based SHM 20 [PITH_FULL_IMAGE:figures/full_fig_p020_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic diagram of the geometry of host plate integrated with surface-bonded PWAS transducers and featuring a notch-type damage [PITH_FULL_IMAGE:figures/full_fig_p021_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Spectral strip element based on the ZIGT, showing 𝑛 physical nodes with four mechanical DoFs and 𝑛𝜙 − 1) internal electric potential DoFs per node and an electric node (p) with 𝑛𝜙 surface electric potential DoFs (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) 1D convolution operation. (b) Max-pooling operation 21 [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Connection between neurons of adjacent layers in a fully-connected layer Hidden layers Input layers Output layers [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Architecture of FFNN 22 [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Activation functions: (a) Linear activation, (b) LeakyRelu activation. Conv1D F=64, K=3,S=1 zero padding MaxPooling1D Pool size =2 MaxPooling1D Pool size =2 Conv1D F=32, K=3,S=1 zero padding Flatten layer Dense Units= 32 Reshape Upsampling size=2 Conv1D F= 64, K=3, S=1 zero padding Conv1D F= 1, K=3, S=1 zero padding Upsampling size=2 Dense Units= 32 Dense Units= Input_shape//4*32 Input Reconstructed output… view at source ↗
Figure 8
Figure 8. Figure 8: Architecture of convolutional autoencoder 23 [PITH_FULL_IMAGE:figures/full_fig_p023_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Hann window-modulated five-cycle tone burst signal of 100 kHz central frequency [PITH_FULL_IMAGE:figures/full_fig_p024_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Mesh configuration in 2D plane strain FE model (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p024_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of 1D TDSE and 2D FE (ABAQUS) solutions for sensory potential for two values of notch depth: (a) 0.5 mm and (b) 2.0 mm. 24 [PITH_FULL_IMAGE:figures/full_fig_p024_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Effect of notch depth on Lamb wave-induced sensor potential (from 1D TDSE ) for notch positions of (a) 𝑑𝐴𝑁 = 70 mm and (b) 𝑑𝐴𝑁 = 85 mm. Epochs Loss [PITH_FULL_IMAGE:figures/full_fig_p025_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Training performance graph for CAE-based TL model 25 [PITH_FULL_IMAGE:figures/full_fig_p025_13.png] view at source ↗
Figure 14
Figure 14. Figure 14 [PITH_FULL_IMAGE:figures/full_fig_p026_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: 1D TDSE solution for sensor potential under (a) 100 kHz excitation and (b) 200 kHz excitation 26 [PITH_FULL_IMAGE:figures/full_fig_p026_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Experimental setup illustrating the arrangement of PZT transducers and acquisition system for Lamb wave generation and sensing. damage damage (1 mm) damage (3 mm) damage (2 mm) damage (3 mm) [PITH_FULL_IMAGE:figures/full_fig_p027_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Configuration of the PZT transducers and block-mass damage positions for generating experimental data 27 [PITH_FULL_IMAGE:figures/full_fig_p027_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Experimental sensor signals for path PZT-1–PZT-3 and damage height of 2 mm under (a) 120 kHz and (b) 150 kHz excitations. Epochs Loss (a) Epochs Loss (b) [PITH_FULL_IMAGE:figures/full_fig_p028_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Training performance of CAE-FFNN-TL model for damage localisation and sizing at (a) pretraining stage with a large 1D TDSE simulation dataset and (b) fine-tuning stage with few experimental data Actual damage size Predicted damage size (a) Actual damage location Predicted damage location (b) [PITH_FULL_IMAGE:figures/full_fig_p028_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Scatter plots of actual and predicted (a) damage size and (b) location from 1D TDSE test data 28 [PITH_FULL_IMAGE:figures/full_fig_p028_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Scatter plots of actual and predicted (a) damage size and (b) location from unseen simulated data Actual damage size Predicted damage size (a) Actual damage location Predicted damage location (b) [PITH_FULL_IMAGE:figures/full_fig_p029_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Scatter plots of actual and predicted (a) damage size and (b) location from limited experimental dataset after transfer learning 29 [PITH_FULL_IMAGE:figures/full_fig_p029_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Scatter plots of actual and predicted (a) damage size and (b) location from unseen experimental data after transfer learning Actual damage size Predicted damage size (a) Actual damage location Predicted damage location (b) [PITH_FULL_IMAGE:figures/full_fig_p030_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: Scatter plots of actual and predicted (a) damage size and (b) location from CAE-FFNN model trained on only limited experimental data 30 [PITH_FULL_IMAGE:figures/full_fig_p030_24.png] view at source ↗
read the original abstract

Guided wave-based structural health monitoring (GWSHM) with onboard transducers offers significant potential for the early diagnosis of damage in engineering structures. However, the practical deployment of deep learning models is often hindered by the limited availability of labelled experimental data and the high computational cost of generating large-scale high-fidelity simulation datasets. This study presents a multifidelity transfer learning framework that integrates lightweight physics-based simulations, convolutional autoencoder (CAE)-based deep feature learning, a feed-forward neural network, and limited experimental measurements for accurate damage localisation and sizing in plate-like structures instrumented with piezoelectric transducers. A computationally efficient one-dimensional time-domain spectral element model is employed to generate a large synthetic dataset for pretraining, while transfer learning adapts the model to experimental domains using only a small amount of labelled data. The CAE-based transfer learning framework significantly outperforms its CNN-based counterpart in damage localisation accuracy. The model achieves excellent predictive performance with $R^2$ scores exceeding 0.93 for damage localisation and 0.99 for damage sizing. Its generalisation capability is demonstrated on previously unseen data, showing high prediction accuracy for damage scenarios not represented during pretraining or fine-tuning. The results establish the proposed framework as an accurate, computationally efficient, and practically viable solution for real-world GWSHM applications.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a multi-fidelity convolutional autoencoder-transfer learning (CAE-TL) framework for guided-wave structural health monitoring. A lightweight 1D time-domain spectral element model generates a large synthetic dataset for CAE pretraining; the learned features are then transferred via fine-tuning on a small set of labeled experimental measurements to predict damage location and size. The authors report that CAE-TL significantly outperforms a CNN baseline, with R² exceeding 0.93 for localization and 0.99 for sizing, and that the model generalizes to previously unseen damage scenarios.

Significance. If the reported transfer performance holds under rigorous cross-validation, the framework would offer a practical route to leverage inexpensive low-fidelity simulations when experimental labels are scarce. The multi-fidelity strategy itself is a constructive response to data limitations in GWSHM; however, the central empirical claims rest on an unexamined 1D-to-2D domain gap whose resolution is not demonstrated in the available description.

major comments (2)
  1. [Abstract] Abstract: the performance claims (R² > 0.93 localization, R² > 0.99 sizing, outperformance over CNN) are presented without any indication of synthetic or experimental dataset cardinalities, the number of labeled experimental samples used for fine-tuning, train/validation/test splits, or uncertainty quantification. These omissions render the numerical results impossible to interpret or reproduce from the given information.
  2. [Abstract / Method] The transfer-learning pipeline (implied in the abstract and method description): the central claim that CAE features pretrained on 1D spectral-element signals transfer effectively to 2D experimental Lamb-wave data after limited fine-tuning is load-bearing, yet the manuscript supplies no analysis of the domain gap. The 1D model necessarily omits lateral spreading, 2D scattering from finite damage, boundary reflections in both in-plane directions, and accurate dispersion surfaces; without explicit evidence (e.g., feature-space alignment metrics or ablation on 2D synthetic data) that these omissions do not prevent useful transfer, the reported generalization cannot be taken as support for the framework.
minor comments (1)
  1. [Abstract] Abstract: the phrase 'previously unseen data' is ambiguous; it should be clarified whether these cases are new experimental measurements or merely held-out synthetic realizations.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We address each major comment below and will revise the manuscript to improve clarity and address the identified gaps where possible.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the performance claims (R² > 0.93 localization, R² > 0.99 sizing, outperformance over CNN) are presented without any indication of synthetic or experimental dataset cardinalities, the number of labeled experimental samples used for fine-tuning, train/validation/test splits, or uncertainty quantification. These omissions render the numerical results impossible to interpret or reproduce from the given information.

    Authors: We agree that the abstract should include these details to ensure the results are interpretable and reproducible. In the revised manuscript, we will expand the abstract to report the cardinalities of the synthetic dataset (generated via the 1D spectral element model) and experimental dataset, the number of labeled experimental samples used for fine-tuning, the train/validation/test splits, and any uncertainty quantification associated with the R² scores. revision: yes

  2. Referee: [Abstract / Method] The transfer-learning pipeline (implied in the abstract and method description): the central claim that CAE features pretrained on 1D spectral-element signals transfer effectively to 2D experimental Lamb-wave data after limited fine-tuning is load-bearing, yet the manuscript supplies no analysis of the domain gap. The 1D model necessarily omits lateral spreading, 2D scattering from finite damage, boundary reflections in both in-plane directions, and accurate dispersion surfaces; without explicit evidence (e.g., feature-space alignment metrics or ablation on 2D synthetic data) that these omissions do not prevent useful transfer, the reported generalization cannot be taken as support for the framework.

    Authors: We acknowledge that the manuscript does not include explicit analyses of the 1D-to-2D domain gap, such as feature-space alignment metrics or ablations using 2D synthetic data. The reported transfer performance on experimental data provides empirical support for the framework, but we agree that additional discussion would strengthen the claims. In the revision, we will add a dedicated discussion of the domain gap, its potential effects, and any supporting evidence derivable from the existing experiments and feature visualizations. revision: partial

Circularity Check

0 steps flagged

No significant circularity; empirical results on held-out data

full rationale

The paper's central claims rest on training a CAE on large synthetic 1D spectral-element data, fine-tuning with limited experimental labels, and reporting R² scores plus generalization on previously unseen experimental test cases. No quoted equations, fitted parameters, or self-citations reduce any reported prediction to a quantity defined by the inputs themselves; the evaluation metrics are standard held-out empirical performance measures. The derivation chain therefore remains self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim depends on the transferability of features learned from low-fidelity simulations to experimental measurements; no new physical entities are introduced.

free parameters (1)
  • CAE and network training hyperparameters
    Architecture choices, learning rates, and layer dimensions are selected and optimized during pretraining and fine-tuning but not enumerated in the abstract.
axioms (1)
  • domain assumption The computationally efficient 1D spectral element model generates data sufficiently representative for feature learning that transfers to experimental conditions
    Invoked when using the synthetic dataset for pretraining the CAE before transfer.

pith-pipeline@v0.9.1-grok · 5769 in / 1177 out tokens · 38478 ms · 2026-06-26T05:05:40.265157+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

78 extracted references

  1. [1]

    Structural Control and Health Monitoring , volume=

    A Deep Transfer Learning Network for Structural Condition Identification with Limited Real-World Training Data , author=. Structural Control and Health Monitoring , volume=. 2023 , publisher=

    2023

  2. [2]

    Kingma, D. P. and Ba, J. , title =. 2014 , eprint =

    2014

  3. [3]

    Smart Materials and Structures , volume=

    Active vibration control of piezoelectric laminated beams with electroded actuators and sensors using an efficient finite element involving an electric node , author=. Smart Materials and Structures , volume=. 2010 , publisher=

    2010

  4. [4]

    Unified efficient layerwise theory for smart beams with segmented extension/shear mode, piezoelectric actuators and sensors , volume =

    Kapuria, Santosh and Hagedorn, Peter , year =. Unified efficient layerwise theory for smart beams with segmented extension/shear mode, piezoelectric actuators and sensors , volume =

  5. [5]

    Boca Raton , volume=

    Introduction to Finite and Spectral Element Methods using MATLAB , author=. Boca Raton , volume=

  6. [6]

    Kapuria, Santosh and Jain, Mayank , journal=. A. 2021 , publisher=

    2021

  7. [7]

    International Journal of Mechanical Sciences , volume=

    Efficient time-domain spectral element with zigzag kinematics for multilayered strips , author=. International Journal of Mechanical Sciences , volume=. 2022 , publisher=

    2022

  8. [8]

    Role of transducer inertia in generation, sensing, and time-reversal process of

    Kapuria, Santosh and Sharma, Bhabagrahi Natha and Arockiarajan, A , journal=. Role of transducer inertia in generation, sensing, and time-reversal process of. 2022 , publisher=

    2022

  9. [9]

    Available at: https://www.sparklerceramics.com/piezoelectricproperties.html , YEAR=

    Sparkler , TITLE=. Available at: https://www.sparklerceramics.com/piezoelectricproperties.html , YEAR=

  10. [10]

    Composite Structures , volume=

    Efficient zigzag theory-based spectral element model for guided waves in composite structures containing delaminations , author=. Composite Structures , volume=. 2023 , publisher=

    2023

  11. [11]

    Deep Learning , author=

  12. [12]

    Journal of Sound and Vibration , volume =

    Efficient electromechanical spectral element model using zigzag kinematics for. Journal of Sound and Vibration , volume =. 2024 , issn =

    2024

  13. [13]

    2014 , publisher=

    Ultrasonic Guided Waves in Solid Media , author=. 2014 , publisher=

    2014

  14. [14]

    Identification of damage using

    Su, Zhongqing and Ye, Lin , journal=. Identification of damage using

  15. [15]

    Giurgiutiu, Victor , journal=. Tuned

  16. [16]

    , journal=

    Su, Zhongqing and Ye, Lin and Lu, Y. , journal=. Guided

  17. [17]

    The use of

    Cawley, Peter , journal=. The use of

  18. [18]

    Philosophical Transactions of the Royal Society A , volume=

    An introduction to structural health monitoring , author=. Philosophical Transactions of the Royal Society A , volume=

  19. [19]

    Philosophical Transactions of the Royal Society A , volume=

    The application of machine learning to structural health monitoring , author=. Philosophical Transactions of the Royal Society A , volume=

  20. [20]

    Computer-Aided Civil and Infrastructure Engineering , volume=

    Structural health monitoring using extremely compressed data through deep learning , author=. Computer-Aided Civil and Infrastructure Engineering , volume=

  21. [21]

    Journal of Manufacturing Systems , volume=

    Deep learning for smart manufacturing: methods and applications , author=. Journal of Manufacturing Systems , volume=

  22. [22]

    Microfabricated

    Luginbuhl, Philippe and Collins, Scott D and Racine, G-A and Gretillat, M-A and De Rooij, Nicolaas F and Brooks, Keith G and Setter, Nava , journal=. Microfabricated. 1997 , publisher=

    1997

  23. [23]

    Journal of Intelligent Material Systems and Structures , volume =

    Jitendra Kumar Agrahari and Santosh Kapuria , title =. Journal of Intelligent Material Systems and Structures , volume =

  24. [24]

    2020 , month =

    Kannusamy, M and Kapuria, S and Sasmal, S , title =. 2020 , month =

    2020

  25. [25]

    Self-focusing and time recompression of

    Ing, RK and Fink, M , journal=. Self-focusing and time recompression of. 1998 , publisher=

    1998

  26. [26]

    Smart Materials and Structures , volume=

    Lamb wave damage severity estimation using ensemble-based machine learning method with separate model network , author=. Smart Materials and Structures , volume=. 2021 , publisher=

    2021

  27. [27]

    Mechanical Systems and Signal Processing , volume=

    Sparse ultrasonic guided wave imaging with compressive sensing and deep learning , author=. Mechanical Systems and Signal Processing , volume=. 2022 , publisher=

    2022

  28. [28]

    Structural Health Monitoring , volume =

    Tong Zhang and Suryakanta Biswal and Ying Wang , title =. Structural Health Monitoring , volume =

  29. [29]

    NDT & E International , volume=

    Defect sizing in guided wave imaging structural health monitoring using convolutional neural networks , author=. NDT & E International , volume=. 2021 , publisher=

    2021

  30. [30]

    Journal of Intelligent Material Systems and Structures , YEAR=

    V Giurgiutiu and B Lin and G Santoni-Bottai and A Cuc , TITLE=. Journal of Intelligent Material Systems and Structures , YEAR=

  31. [31]

    Structural Control and Health Monitoring , year=

    Wang, Peng and Zhou, Wensong and Bao, Yuequan and Li, Hui , title=. Structural Control and Health Monitoring , year=

  32. [32]

    Sensors , year=

    Hu, Chaojie and Yang, Bin and Xuan, Fu-Zhen and Yan, Jianjun and Xiang, Yanxun , title=. Sensors , year=

  33. [33]

    Ultrasonics , year=

    Huan, Qiang and Chen, Mingtong and Li, Faxin , title=. Ultrasonics , year=

  34. [34]

    Identification of Damage using Lamb Waves

    Z Su and Lin Ye. Identification of Damage using Lamb Waves. 2009

    2009

  35. [35]

    Smart Materials and Structures , volume=

    Lamb wave-based baseline-free damage localization in pipes with surface-mounted piezo patches , author=. Smart Materials and Structures , volume=. 2025 , publisher=

    2025

  36. [36]

    Composite Structures , pages=

    Autonomous structural health monitoring of composite wind turbine blades using guided waves and machine learning , author=. Composite Structures , pages=. 2025 , publisher=

    2025

  37. [37]

    Journal of Nondestructive Evaluation , YEAR=

    R P Dalton and P Cawley and M J S Lowe , TITLE=. Journal of Nondestructive Evaluation , YEAR=

  38. [38]

    Mechanical Systems and Signal Processing , volume=

    Isolation of ultrasonic scattering by wavefield baseline subtraction , author=. Mechanical Systems and Signal Processing , volume=. 2016 , publisher=

    2016

  39. [39]

    Structural Health Monitoring system based on a concept of

    Kudela, Pawel and Radzienski, Maciej and Ostachowicz, Wieslaw and Yang, Zhibo , journal=. Structural Health Monitoring system based on a concept of. 2018 , publisher=

    2018

  40. [40]

    Wave motion , volume=

    Guided wave signal processing and image fusion for in situ damage localization in plates , author=. Wave motion , volume=. 2007 , publisher=

    2007

  41. [41]

    Mechanical Systems and Signal Processing , volume=

    Effective combination of modeling and experimental data with deep metric learning for guided wave-based damage localization in plates , author=. Mechanical Systems and Signal Processing , volume=. 2022 , publisher=

    2022

  42. [42]

    Composite Structures , volume=

    An integrated multi-task transfer learning for damage detection, localization, and severity assessment of laminated composite plate , author=. Composite Structures , volume=. 2025 , publisher=

    2025

  43. [43]

    Best reconstruction frequency for time-reversal process of

    Kapuria, Santosh , journal=. Best reconstruction frequency for time-reversal process of. 2025 , publisher=

    2025

  44. [44]

    Physics-guided neural network for structural health monitoring with

    Song, Yang and Shan, Shengbo and Zhang, Yuanman and Cheng, Li , journal=. Physics-guided neural network for structural health monitoring with. 2026 , publisher=

    2026

  45. [45]

    Thermal sensitivity of

    Dodson, JC and Inman, DJ , journal=. Thermal sensitivity of. 2013 , publisher=

    2013

  46. [46]

    Baseline-free damage detection and sizing under varying temperatures using

    Sharma, Bhabagrahi Natha and Kapuria, Santosh and Arockiarajan, A and Kannusamy, M , journal=. Baseline-free damage detection and sizing under varying temperatures using. 2023 , publisher=

    2023

  47. [47]

    Time reversibility of

    Sharma, Bhabagrahi Natha and Kapuria, Santosh and Arockiarajan, A , journal=. Time reversibility of. 2021 , publisher=

    2021

  48. [48]

    NDT & E International , volume=

    Guided waves in long range nondestructive testing and structural health monitoring: Principles, history of applications and prospects , author=. NDT & E International , volume=. 2024 , publisher=

    2024

  49. [49]

    Ultrasonics , volume=

    Appraisal of linear baseline-free techniques for guided wave based structural health monitoring , author=. Ultrasonics , volume=. 2024 , publisher=

    2024

  50. [50]

    Smart Materials and Structures , volume=

    Systematic critical review of structural health monitoring under environmental and operational variability: approaches for baseline compensation, adaptation, and reference-free techniques , author=. Smart Materials and Structures , volume=. 2025 , publisher=

    2025

  51. [51]

    Mechanical Systems and Signal Processing , volume =

    Fast and accurate decomposition of overlapping. Mechanical Systems and Signal Processing , volume =. 2026 , author =

    2026

  52. [52]

    Ultrasonics , volume=

    A review on guided-ultrasonic-wave-based structural health monitoring: From fundamental theory to machine learning techniques , author=. Ultrasonics , volume=. 2023 , publisher=

    2023

  53. [53]

    Deep learning based crack damage detection technique for thin plate structures using guided

    Liu, Heng and Zhang, Yunfeng , journal=. Deep learning based crack damage detection technique for thin plate structures using guided. 2020 , publisher=

    2020

  54. [54]

    DeepSHM: A deep learning approach for structural health monitoring based on guided

    Ewald, Vincentius and Groves, Roger M and Benedictus, Rinze , booktitle=. DeepSHM: A deep learning approach for structural health monitoring based on guided. 2019 , organization=

    2019

  55. [55]

    Composite Structures , volume=

    Lamb wave-based damage detection of composite structures using deep convolutional neural network and continuous wavelet transform , author=. Composite Structures , volume=. 2021 , publisher=

    2021

  56. [56]

    Smart Materials and Structures , volume=

    Lamb wave based damage detection in metallic plates using multi-headed 1-dimensional convolutional neural network , author=. Smart Materials and Structures , volume=. 2021 , publisher=

    2021

  57. [57]

    Mechanical Systems and Signal Processing , volume=

    Damage localization in plate-like structures using time-varying feature and one-dimensional convolutional neural network , author=. Mechanical Systems and Signal Processing , volume=. 2021 , publisher=

    2021

  58. [58]

    Mechanical Systems and Signal Processing , volume=

    On the explainability of convolutional neural networks processing ultrasonic guided waves for damage diagnosis , author=. Mechanical Systems and Signal Processing , volume=. 2023 , publisher=

    2023

  59. [59]

    Explainable

    Pandey, Pushpa and Rai, Akshay and Mitra, Mira , journal=. Explainable. 2022 , publisher=

    2022

  60. [60]

    IEEE transactions on neural networks , volume=

    An overview of statistical learning theory , author=. IEEE transactions on neural networks , volume=. 1999 , publisher=

    1999

  61. [61]

    A numerically-enhanced machine learning approach to damage diagnosis using a

    Sbarufatti, Claudio and Manson, G and Worden, K , journal=. A numerically-enhanced machine learning approach to damage diagnosis using a. 2014 , publisher=

    2014

  62. [62]

    Application of Artificial Neural Networks and Probability Ellipse methods for damage detection using

    De Fenza, Angelo and Sorrentino, Assunta and Vitiello, Pasquale , journal=. Application of Artificial Neural Networks and Probability Ellipse methods for damage detection using. 2015 , publisher=

    2015

  63. [63]

    Computers & Structures , volume=

    Crack detection in Mindlin-Reissner plates under dynamic loads based on fusion of data and models , author=. Computers & Structures , volume=. 2021 , publisher=

    2021

  64. [64]

    Journal of Mechanical Design , volume=

    Calibration and validation framework for selective laser melting process based on multi-fidelity models and limited experiment data , author=. Journal of Mechanical Design , volume=. 2020 , publisher=

    2020

  65. [65]

    Ultrasonics , volume=

    A physics-embedded deep-learning framework for efficient multi-fidelity modeling applied to guided wave based structural health monitoring , author=. Ultrasonics , volume=. 2024 , publisher=

    2024

  66. [66]

    Journal of Big data , volume=

    A survey of transfer learning , author=. Journal of Big data , volume=. 2016 , publisher=

    2016

  67. [67]

    Proceedings of the IEEE , volume=

    A comprehensive survey on transfer learning , author=. Proceedings of the IEEE , volume=. 2020 , publisher=

    2020

  68. [68]

    Computer-aided civil and Infrastructure Engineering , volume=

    Dynamics-based cross-domain structural damage detection through deep transfer learning , author=. Computer-aided civil and Infrastructure Engineering , volume=. 2022 , publisher=

    2022

  69. [69]

    Computer Methods in Applied Mechanics and Engineering , volume=

    Sustainable computational mechanics assisted by deep learning , author=. Computer Methods in Applied Mechanics and Engineering , volume=. 2022 , publisher=

    2022

  70. [70]

    A variable kinematic multi-field model for

    Najd, Jamal and Zappino, Enrico and Carrera, Erasmo and Harizi, Walid and Aboura, Zoheir , journal=. A variable kinematic multi-field model for. 2025 , publisher=

    2025

  71. [71]

    Mechanics of Advanced Materials and Structures , volume=

    Finite element modeling and analysis of adhesive layer effects in surface-bonded piezoelectric sensors and actuators including non-uniform thickness , author=. Mechanics of Advanced Materials and Structures , volume=. 2022 , publisher=

    2022

  72. [72]

    International Journal of Mechanical Sciences , volume=

    A coupled efficient layerwise finite element model for free vibration analysis of smart piezo-bonded laminated shells featuring delaminations and transducer debonding , author=. International Journal of Mechanical Sciences , volume=. 2021 , publisher=

    2021

  73. [73]

    Efficient layerwise multiphysics spectral element model for delaminated composite strips with

    Jain, Mayank and Kapuria, Santosh , journal=. Efficient layerwise multiphysics spectral element model for delaminated composite strips with. 2025 , publisher=

    2025

  74. [74]

    Journal of Sound and Vibration , volume=

    Spectral finite element based on an efficient layerwise theory for wave propagation analysis of composite and sandwich beams , author=. Journal of Sound and Vibration , volume=. 2014 , publisher=

    2014

  75. [75]

    Journal of Intelligent Material Systems and Structures , volume=

    A time domain spectral layerwise finite element for wave structural health monitoring in composite strips with physically modeled active piezoelectric actuators and sensors , author=. Journal of Intelligent Material Systems and Structures , volume=. 2017 , publisher=

    2017

  76. [76]

    Aircraft Engineering and Aerospace Technology , volume=

    Spectral finite element method for wave propagation analysis in smart composite beams containing delamination , author=. Aircraft Engineering and Aerospace Technology , volume=. 2020 , publisher=

    2020

  77. [77]

    International Journal of Mechanical Sciences , volume=

    A wave packet enriched finite element for electroelastic wave propagation problems , author=. International Journal of Mechanical Sciences , volume=. 2020 , publisher=

    2020

  78. [78]

    2014 , publisher=

    Structural health monitoring with piezoelectric wafer active sensors , author=. 2014 , publisher=

    2014