Skip to main content
SpringerLink
Log in

Delamination Diagnosis System Using Nonlinear Transformation-Based Augmentation Approach for CNN Transfer Learning

  • Original Paper
  • Published:
Journal of Vibration Engineering & Technologies Aims and scope Submit manuscript

Abstract

Purpose

This study presents a nonlinear transformation-based data augmentation approach to diagnose delaminations such as interlaminar, through-hole and buckle, in relatively complex composites. The approach uses tailored vibration signals of relatively simplified composites and employs a convolutional neural network (CNN).

Methods

The present study involves modeling several composite structures with and without interlaminar, through-hole and buckle delaminations using finite-element method (FEM)-based simulation. A vibration experiment is also conducted for the case of interlaminar delamination. In our method, it is assumed that all vibration signals of simplified composites and healthy condition of complex real composites are obtained that paves the way to investigate the delamination conditions of real composites. Then, the present approach adopts the nonlinear transformation method mapping the signals of complex healthy composites and simplified healthy composites. The difference between the transformed signals of complex real composites and the reference signals of simplified composites is utilized for the virtual spectrogram computed by the short-time Fourier Transform (STFT). These virtual spectrograms fed into a CNN-based diagnosis system to diagnose the types and locations of delaminations.

Results

For interlaminar cases, the simulation and experimental vibration signals are diagnosed with 100 and 92.5%, respectively. Through-hole and buckle cases are both diagnosed with an accuracy of 100%. Using nonlinear transformation to classify vibration signals prior to diagnosis has been shown to be highly advantageous for delamination diagnosis.

Conclusion

The results allow the diagnosis of three cases of complex composites with simplified composites using the delamination diagnosis system.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15

Similar content being viewed by others

Data Availibility

Data are available from the corresponding author on reasonable request.

References

  1. Meng M, Chua YJ, Wouterson E, Ong CP, Kelvin (2017) Ultrasonic signal classification and imaging system for composite materials via deep convolutional neural networks. Neurocomputing 257:128–135

  2. Barile C, Casavola C, Pappalettera G, Kannan VP (2022) Damage monitoring of carbon fibre reinforced polymer composites using acoustic emission technique and deep learning. Compos Struct 292:115629

    Google Scholar 

  3. Scholz V, Winkler P, Hornig A, Gude M, Filippatos A (2021) Structural damage identification of composite rotors based on fully connected neural networks and convolutional neural networks. Sensors 21(6):2005

    Google Scholar 

  4. Shorten C, Khoshgoftaar TM (2019) A survey on image data augmentation for deep learning. Journal of Big Data 6(1):1–48

    Google Scholar 

  5. Wang J, Perez L et al (2017) The effectiveness of data augmentation in image classification using deep learning. Convolutional Neural Networks Vis. Recognit 11:1–8

    Google Scholar 

  6. Cubuk Ekin D, Zoph Barret, Mane Dandelion, Vasudevan Vijay, Le Quoc V (2019) Autoaugment: Learning augmentation strategies from data. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 113-123

  7. Khan A, Raouf I, Noh YR, Lee D, Sohn JW, Kim HS (2022) Autonomous assessment of delamination in laminated composites using deep learning and data augmentation. Compos Struct 209:115502

    Google Scholar 

  8. Dabetwar Shweta, Ekwaro-Osire Stephen, Dias João Paulo (2022) Fatigue damage diagnostics of composites using data fusion and data augmentation with deep neural networks. Journal of Nondestructive Evaluation, Diagnostics and Prognostics of Engineering Systems 5(2)

  9. Hussain Mahbub, Bird Jordan J, Faria Diego R (2018) A study on cnn transfer learning for image classification. UK Workshop on computational Intelligence 191-202

  10. Shin H-C, Roth HR, Gao M, Lu L, Xu Z, Nogues I, Yao J, Mollura D, Summers RM (2016) Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning. IEEE Trans Med Imaging 35(5):1285–1298

    Google Scholar 

  11. Han D, Liu Q, Fan W (2018) A new image classification method using CNN transfer learning and web data augmentation. Expert Syst Appl 95:43–56

    Google Scholar 

  12. Zou Y, Tong LPSG, Steven GP (2000) Vibration-based model-dependent damage (delamination) identification and health monitoring for composite structures-a review. J Sound Vib 230(2):357–378

    Google Scholar 

  13. Zhang Z, Shankar K, Ray T, Morozov EV, Tahtali M (2013) Vibration-based inverse algorithms for detection of delamination in composites. Compos Struct 102:226–236

    Google Scholar 

  14. Zhang Z, Shankar K, Morozov EV, Tahtali M (2016) Vibration-based delamination detection in composite beams through frequency changes. J Vib Control 22(2):496–512

    Google Scholar 

  15. Yam LH, Wei Z, Cheng L (2004) Nondestructive detection of internal delamination by vibration-based method for composite plates. J Compos Mater 38(24):2183–2198

    Google Scholar 

  16. Jakkamputi L, Devaraj S, Marikkannan S, Gnanasekaran S, Ramasamy S, Rakkiyannan J, Xu Y (2022) Experimental and Computational Vibration Analysis for Diagnosing the Defects in High Performance Composite Structures Using Machine Learning Approach. Appl Sci 12(23):12100

    Google Scholar 

  17. Mei H, Migot A, Haider MF, Joseph R, Yeasin BM, Giurgiutiu V (2019) Vibration-based in-situ detection and quantification of delamination in composite plates. Sensors 19(7):1734

    Google Scholar 

  18. Migot Asaad, Giurgiutiu Victor (2022) Numerical and experimental investigation of delamination severity estimation using local vibration techniques. Journal of Intelligent Material Systems and Structures 1045389X221128585

  19. Park G, Rutherford AC, Wait JR, Nadler B, Farrar C, Claytor TN (2005) High-frequency response functions for composite plate monitoring with ultrasonic validation. AIAA J 43(11):2431–2437

    Google Scholar 

  20. Lim DK, Mustapha KB, Pagwiwoko CP (2021) Delamination detection in composite plates using random forests. Compos Struct 278:114676

    Google Scholar 

  21. Khan A, Ko D-K, Lim SC, Kim HS (2019) Structural vibration-based classification and prediction of delamination in smart composite laminates using deep learning neural network. Compos B Eng 161:586–594

    Google Scholar 

  22. Khan A, Shin JK, Lim WC, Kim NY, Kim HS (2020) A deep learning framework for vibration-based assessment of delamination in smart composite laminates. Sensors 20(8):2335

    Google Scholar 

  23. Saadatmorad M, Jafari-Talookolaei R-A, Pashaei M-H, Khatir S (2022) Damage Detection in Rectangular Laminated Composite Plate Structures using a Combination of Wavelet Transforms and Artificial Neural Networks. Journal of Vibration Engineering & Technologies 10(5):1647–1664

    Google Scholar 

  24. Jena PC, Parhi DR, Pohit G (2019) Dynamic Investigation of FRP cracked beam using neural network technique. Journal of Vibration Engineering & Technologies 7:647–661

    Google Scholar 

  25. Dabetwar Shweta, Ekwaro-Osire Stephen, Dias João Paulo (2022) Fatigue damage diagnostics of composites using data fusion and data augmentation with deep neural networks. Journal of Nondestructive Evaluation, Diagnostics and Prognostics of Engineering Systems 5(2)

  26. Khan A, Kim HS (2022) A Brief Overview of Delamination Localization in Laminated Composites. Multiscale Science and Engineering 4(3):102–110

    Google Scholar 

  27. Weiss K, Khoshgoftaar TM, Wang DD (2016) A survey of transfer learning. Journal of Big data 3(1):1–40

    Google Scholar 

  28. Goswami S, Anitescu C, Chakraborty S, Rabczuk T (2020) Transfer learning enhanced physics informed neural network for phase-field modeling of fracture. Theoret Appl Fract Mech 106:102447

    Google Scholar 

  29. Kathirvel P, Manikandan MS, Prasanna SRM, Soman KP (2011) An efficient R-peak detection based on new nonlinear transformation and first-order Gaussian differentiator. Cardiovasc Eng Technol 2(4):408–425

    Google Scholar 

  30. Maragos P (1995) Slope transforms: theory and application to nonlinear signal processing. IEEE Trans Signal Process 43(4):864–877

    Google Scholar 

  31. Sun L, Hou J, Xing C, Fang Z (2022) A robust hammerstein-wiener model identification method for highly nonlinear systems. Processes 10(12):2664

    Google Scholar 

  32. Lu Z-Q, Gu D-H, Ding H, Lacarbonara W, Chen L-Q (2020) Nonlinear vibration isolation via a circular ring. Mech Syst Signal Process 136:106490

    Google Scholar 

  33. Xiao X, Zhang Q, Zheng J, Li Z (2023) Analytical model for the nonlinear buckling responses of the confined polyhedral FGP-GPLs lining subjected to crown point loading. Eng Struct 282:115780

    Google Scholar 

  34. Shang K, Chen Z, Liu Z, Song L, Zheng W, Yang B, Liu S, Yin L (2021) Haze prediction model using deep recurrent neural network. Atmosphere 12(12):1625

    Google Scholar 

  35. Yin L, Wang L, Huang W, Tian J, Liu S, Yang B, Zheng W (2022) Haze grading using the convolutional neural networks. Atmosphere 13(4):522

    Google Scholar 

  36. Lv Z, Wu J, Li Y, Song H (2022) Cross-layer optimization for industrial Internet of Things in real scene digital twins. IEEE Internet Things J 9(17):15618–15629

    Google Scholar 

  37. Xu S, Dai H, Feng L, Chen H, Chai Y, Zheng WX (2023) Fault Estimation for Switched Interconnected Nonlinear Systems with External Disturbances via Variable Weighted Iterative Learning. Express Briefs, IEEE Transactions on Circuits and Systems II

    Google Scholar 

  38. Zhan Chuanjun, Dai Zhenxue, Yang Zhijie, Zhang Xiaoying, Ma Ziqi, Thanh Hung Vo, Soltanian Mohamad Reza (2023) Subsurface sedimentary structure identification using deep learning: A review. Earth-Science Reviews 104370

  39. Zhang Xu, Huang Dengbing, Li Hanyu, Zhang Youjia, Xia Ying, Liu Jinzhuo (2023) Self–training maximum classifier discrepancy for EEG emotion recognition. CAAI Transactions on Intelligence Technology

  40. Wang J, Wu D, Gao Y, Wang X, Li X, Xu G, Dong W (2022) Integral real-time locomotion mode recognition based on GA-CNN for lower limb exoskeleton. J Bionic Eng 19(5):1359–1373

    Google Scholar 

  41. Dang Wei, Xiang Longhai, Liu Shan, Yang Bo, Liu Mingzhe, Yin Zhengtong, Yin Lirong, Zheng Wenfeng (2023) A Feature Matching Method based on the Convolutional Neural Network. Journal of Imaging Science and Technology 1-11

  42. Wang H, Gao Q, Li H, Wang H, Yan L, Liu G (2022) A structural evolution-based anomaly detection method for generalized evolving social networks. Comput J 65(5):1189–1199

    Google Scholar 

  43. Liu H, Xu Y, Chen F (2023) Sketch2Photo: Synthesizing photo-realistic images from sketches via global contexts. Eng Appl Artif Intell 117:105608

    Google Scholar 

  44. Liu H, Liu M, Li D, Zheng W, Yin L, Wang R (2022) Recent advances in pulse-coupled neural networks with applications in image processing. Electronics 11(20):3264

    Google Scholar 

  45. Zhou G, Song B, Liang P, Xu J, Yue T (2022) Voids filling of DEM with multiattention generative adversarial network model. Remote Sensing 14(5):1206

    Google Scholar 

  46. Liu C, Cui J, Zhang Z, Liu H, Huang X, Zhang C (2021) The role of TBM asymmetric tail-grouting on surface settlement in coarse-grained soils of urban area: Field tests and FEA modelling. Tunn Undergr Space Technol 111:103857

    Google Scholar 

  47. Turan Muhittin, Yaylacı Uzun, Ecren and Yaylacı, Murat, (2023) Free vibration and buckling of functionally graded porous beams using analytical, finite element, and artificial neural network methods. Archive of Applied Mechanics 93(4):1351–1372

  48. Liu J, Wang L (2023) Two-stage vibration-suppression framework for optimal robust placements design and reliable PID gains design via set-crossing theory and artificial neural network. Reliability Engineering & System Safety 230:108956

    Google Scholar 

  49. Avcar M, Saplioglu K (2015) An artificial neural network application for estimation of natural frequencies of beams. Int J Adv Comput Sci Appl 6(6):94–102

    Google Scholar 

  50. Civalek Ö, Avcar M (2022) Free vibration and buckling analyses of CNT reinforced laminated non-rectangular plates by discrete singular convolution method. Engineering with Computers 38(Suppl 1):489–521

    Google Scholar 

  51. Yaylaci M, Yaylaci EU, Ozdemir ME, Ozturk Ş, Sesli H (2023) Vibration and buckling analyses of FGM beam with edge crack: Finite element and multilayer perceptron methods. Steel Compos Struct 46(4):565

    Google Scholar 

  52. Liu Y, Wang L, Li M, Wu Z (2022) A distributed dynamic load identification method based on the hierarchical-clustering-oriented radial basis function framework using acceleration signals under convex-fuzzy hybrid uncertainties. Mech Syst Signal Process 172:108935

    Google Scholar 

  53. Grassi M, Zhang X (2003) Finite element analyses of mode I interlaminar delamination in z-fibre reinforced composite laminates. Compos Sci Technol 63(12):1815–1832

    Google Scholar 

  54. Hashemi S, Kinloch Anthony James, Williams JM (1990) The analysis of interlaminar fracture in uniaxial fibre-polymer composites. Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences 427(1872):173-199

  55. Jahanian E, Zeinedini A (2018) Influence of drilling on mode II delamination of E-glass/epoxy laminated composites. Theoret Appl Fract Mech 96:398–407

    Google Scholar 

  56. Hou JP, Petrinic N, Ruiz C (2001) A delamination criterion for laminated composites under low-velocity impact. Compos Sci Technol 61(14):2069–2074

    Google Scholar 

  57. Hwang S-F, Liu G-H (2001) Buckling behavior of composite laminates with multiple delaminations under uniaxial compression. Compos Struct 53(2):235–243

    Google Scholar 

  58. Short GJ, Guild FJ, Pavier MJ (2001) The effect of delamination geometry on the compressive failure of composite laminates. Compos Sci Technol 61(14):2075–2086

    Google Scholar 

  59. Wisnom MR (2012) The role of delamination in failure of fibre-reinforced composites. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 370(1965):1850–1870

    Google Scholar 

  60. Kim D-Y, Woo Y-J, Kang K, Yoon GH (2022) Failure diagnosis system using a new nonlinear mapping augmentation approach for deep learning algorithm. Mech Syst Signal Process 172:108914

    Google Scholar 

  61. Harris FJ (1978) On the use of windows for harmonic analysis with the discrete Fourier transform. Proc IEEE 66(1):51–83

    Google Scholar 

  62. Randall RB (1987) Frequency analysis. Bruel & Kjaer, Naerum (DK)

    Google Scholar 

  63. Kamilaris A, Prenafeta-Boldú FX (2018) Deep learning in agriculture: A survey. Comput Electron Agric 147:70–90

    Google Scholar 

  64. Haidong S, Hongkai J, Xingqiu L, Shuaipeng W (2018) Intelligent fault diagnosis of rolling bearing using deep wavelet auto-encoder with extreme learning machine. Knowl-Based Syst 140:1–14

    Google Scholar 

  65. Chen Z, Zhang T, Ouyang C (2018) End-to-end airplane detection using transfer learning in remote sensing images. Remote Sensing 10(1):139

    Google Scholar 

  66. Shen Zejiang, Wan Xili, Ye Feng, Guan Xinjie, Liu Shuwen (2019) Deep learning based framework for automatic damage detection in aircraft engine borescope inspection. 2019 International Conference on Computing, Networking and Communications (ICNC) 1005-1010

  67. Liang G, Zheng L (2020) A transfer learning method with deep residual network for pediatric pneumonia diagnosis. Comput Methods Programs Biomed 187:104964

    Google Scholar 

  68. Krois J, Ekert T, Meinhold L, Golla T, Kharbot B, Wittemeier A, Dörfer C, Schwendicke F (2019) Deep learning for the radiographic detection of periodontal bone loss. Sci Rep 9(1):1–6

    Google Scholar 

  69. Gong Y, Shao H, Luo J, Li Z (2020) A deep transfer learning model for inclusion defect detection of aeronautics composite materials. Compos Struct 252:112681

    Google Scholar 

  70. MATLAB (2021) Deep Learning Toolbox version 14.2 (R2021a). Natick, Massachusetts: The MathWorks Inc

Download references

Funding

The authors are grateful for financial support received from National Research Foundation (NRF) of the Korea grant funded by Korea government (MSIT) (No. 2018R1A5A7025522).

Author information

Authors and Affiliations

  1. School of Mechanical Engineering, Hanyang University, Seoul, 04763, South Korea

    Dong-Yoon Kim, Yeon-Jun Woo, Seong-Gyu Sim & Gil Ho Yoon

Authors
  1. Dong-Yoon Kim
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Yeon-Jun Woo
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Seong-Gyu Sim
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Gil Ho Yoon
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Gil Ho Yoon.

Ethics declarations

Conflict of Interest

The authors declare that they have no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

Detail Geometries

In this present study, various geometries are employed to conduct several examples. These geometries were applied identically to the finite-element models. Table 1 shows the detailed geometries of the finite-element models. In the experiment, data were generated by repeating the experiment for each single geometry specimen. Table 2 shows the detailed geometries of the experiment specimens.

Table 1 Detail geometries of the finite-element models
Full size table
Table 2 Detail geometries of the experiment specimens
Full size table

Auxiliary Example 1: Vertical Through-Hole

For the auxiliary example, the through-hole case is investigated in Fig. 16. The same material properties are set except for the through-hole. The responses of the through-holes with 5 mm diameter at the three different locations (57.5 mm, 115 mm and 172.5 mm distances from the end of the right side) are computed. The three different holes of complex delaminated models are modeled differently from simplified delaminated models and located at 100 mm, 200 mm and 300 mm distances from the end of the right side. The boundary condition including the impact loading is set to the same of the first example. The responses of the simplified reference models (H.R.S.) with and without the through-hole are easily obtained and the response of the complex healthy model (H.S.) is obtained prior. The nonlinear transformation function can be defined in Fig 17a. Then, the response of the unknown real system is measured and mapped in Fig. 17b. This transformed signal is then defined as the mapped unknown signal (M.U.S.) in Fig. 17b. By comparing the simplified reference signals and mapped complex signals, the condition of the unknown model can be approximately determined. The training and verification data were decided with the virtual spectrograms generated by the differences between the vibration signals of the simplified and complex models in Fig. 18. In the diagnosis system, Fig. 9a shows that it is possible to diagnose the through-hole and health conditions with high accuracy and the diagnosis system that applied the concept of transfer learning using CNN was effective.

Fig. 16
figure 16

Auxiliary example 1: Illustration of geometrical configurations about several simplified reference and complex real models with vertical through holes

Full size image
Fig. 17
figure 17

The procedure of nonlinear transformation with through hole cases. a Transformation process of the FRFs of simplified reference and complex healthy models and b transformation process of the FRFs of simplified reference and complex models

Full size image
Fig. 18
figure 18

The virtual spectrograms of representative models. a The virtual spectrograms to be utilized for training with and without the delamination and b the virtual spectrograms to be utilized for testing with the locations of through hole at 100 mm, 200 mm and 300 mm

Full size image

Auxiliary Example 2: Buckle Delamination with Compressive Force

For the auxiliary example, a monitoring problem for buckle delamination which is one of the most widely observed failures especially for thin film or coating on a substrate is considered in Fig. 19. All the detailed parameters are presented in the figures. In Figs. 19, Fig. 20, Fig. 21 and Fig. 9b, the results show that the present scheme can be applied for the buckle delamination too.

Finally, the diagnosis system is performed once more to diagnose all delamination types (interlaminar, through-hole and buckle delamination cases) and locations. To carry out the diagnosis, 5 data are selected for all cases and are used for training and verification data. As a result, it was possible to make a diagnosis system to classify various delamination types and locations successfully with 100% accuracy in Fig. 10. Although the diagnosis of delaminations using the FE model is performed well, the diagnosis may not be performed with high accuracy in a real experiment.

Fig. 19
figure 19

Auxiliary example 2: Illustration of geometrical configurations about several simplified reference and complex real models with buckle delamination

Full size image
Fig. 20
figure 20

The procedure of nonlinear transformation with buckle delamination cases. a Transformation process of the FRFs of simplified reference and complex healthy models and b transformation process of the FRFs of simplified reference and complex models

Full size image
Fig. 21
figure 21

The virtual spectrograms of representative models. a The virtual spectrograms to be utilized for training with and without the delamination and b the virtual spectrograms to be utilized for testing with the locations of buckle delamination at 110 mm, 200 mm and 290 mm

Full size image

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kim, DY., Woo, YJ., Sim, SG. et al. Delamination Diagnosis System Using Nonlinear Transformation-Based Augmentation Approach for CNN Transfer Learning. J. Vib. Eng. Technol. 12, 3213–3230 (2024). https://doi.org/10.1007/s42417-023-01040-1

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s42417-023-01040-1

Keywords

Search

Navigation