Nondestructive evaluation of carbon-fiber composites using digital image correlation,acoustic emission,and optical based modal analysis

Abstract

Composite materials are increasingly used in the wind industries. Damage detection and health monitoring of composite materials are challenging due to the complex internal structure and unique material properties. Digital image correlation (DIC) and acoustic emission (AE) are both used for damage detection in structures. In this work, DIC performs a full-field strain measurement on the surface of the carbon-fiber specimen while AE continuously monitors and records the AE signals generated from specimen subsurface structure failures. These health monitoring techniques are integrated and evaluated in this study to correlate surface strain measurements and acoustic emission measurements on carbon-fiber specimens. The AE measurement results show that there is a correlation between the occurrence of AE events and the timing of complete specimen failure. DIC with a high-speed stereo camera system is also adopted to extract the change in the resonance frequencies and displacement and strain mode shapes of the specimen during experiments in cyclic loading.

Keywords

Digital image correlation (DIC)acoustic emission (AE)carbon-fiber failure pattern vibrations structural health monitoring

Introduction

Composite materials are increasingly used in aerospace structures (Diamanti and Soutis, 2010) and wind turbines (Brøndsted et al., 2005; Poozesh et al., 2017) due to their high stiffness and strength, low weight, and low thermal expansion. In the aerospace industry, modern aircraft continue to use more composite materials in load-carrying structures and turbine blades of the jet engine, such as GE9X, to reduce aircraft weight (Diamanti and Soutis, 2010), achieve better fuel economy, and meet more stringent carbon dioxide emission regulations. In the wind turbine industry, modern wind turbines continue to grow in size to help reduce the cost of energy capture (Fingersh et al., 2006; Islam et al., 2013). However, as the size of the wind turbines scales up, the capital investment for these machines is escalating. In order to maintain the structural integrity and crucial safety of the aircraft and to reduce wind turbine operational cost, the composite structure health has to be inspected and continuously monitored without the disruption of the structure’s operation. However, damage detection and health monitoring of composite materials are quite challenging due to the complex internal structure and unique material properties. Therefore, non-contact and nondestructive inspection-based monitoring of these structures has become of particular interest.

In the field of structural health monitoring, many non-contact and nondestructive inspection techniques have been adopted to inspect structural integrity, including visual inspection, shearography or speckle pattern shearing interferometry, ultrasonic inspection, vibration analysis, radiography, thermography, acoustic emission (AE), and digital image correlation (DIC) (Ida and Meyendorf, 2019; Niezrecki et al., 2019). AE is generated by acoustic waves which are produced from irreversible changes such as cracks or dislocations within the material (Grosse and Ohtsu, 2008). While most methods detect geometrical discontinuities, recent research has demonstrated that AE can detect different types of internal material damages, such as fiber breakage, fiber pull-out, matrix cracking, and delamination inside laminated composite plates (Diamanti and Soutis, 2010; McCrory et al., 2015). With multiple sensors and careful analysis, it can also be possible to locate the sources of the AE events (Eaton et al., 2012; McCrory et al., 2015). However, acoustic emission cannot measure the severity of the damage, and it is susceptible to noise from the environment. Recently, non-contact optical methods for rapid inspection over the full structure have received more attention due to their high spatial resolution, high precision, and lower instrumentation costs. Of these various “full-field” methods, DIC (Baqersad et al., 2017; Goidescu et al., 2013) and three-dimensional point tracking (Baqersad et al., 2015a) have both recently been studied for the health monitoring of wind turbine blades (Niezrecki et al., 2019). Recent studies show that 3D DIC can effectively monitor the integrity of the civil infrastructure (Cigada et al., 2013) or biological systems (Mostafavi Yazdi and Baqersad, 2022; Panchal et al., 2019). DIC cameras can be installed at a distance (Ozbek et al., 2013) or mounted on a drone (Khadka et al., 2020, 2022) to be used for in-situ measurement of utility-scale wind turbines. However, the measured results from the DIC approach usually need to be post-processed and combined with other health monitoring algorithms to find the defects in the structures (Baqersad et al., 2015a; Chen et al., 2018), and DIC is only capable of identifying surface strain and cracks. However, there are few to no papers that use DIC along with AE. Also, using DIC strain mode shapes for structural health monitoring has not been studied in the literature.

In this paper, digital image correlation (DIC) and acoustic emission (AE) techniques are evaluated for the structural health monitoring of carbon-fiber laminates under a tensile loading test. The full-field surface strain of the carbon-fiber specimen obtained using DIC combined with subsurface AE signals obtained with the acoustic emission is used to predict the location and timing of the complete failure of the carbon-fiber specimen. Furthermore, the work uses DIC to obtain vibration mode shapes and evaluate how vibration properties of composites can be used for health monitoring. This is the first time that strain mode shapes of a structure obtained with DIC are used for damage detection.

Methodology

DIC, AE, and modal analysis techniques were all adopted in this study, and their performances for structural health monitoring of carbon-fiber laminate specimens were evaluated. Figure 1 shows the overall systematic approach investigated in this study. In the static tensile strength experiment, the specimens were subjected to a controlled static tensile loading rate with real-time structure health monitoring, including DIC and AE measurements. DIC can capture the full-field strain measurements of specimens and can identify the high strain area on the specimen surface, where the complete specimen failure occurs. Moreover, AE equipment can monitor the specimen’s sub-surface defects through the signals from the AE hits, and it can also locate major AE hits in the longitudinal direction with the help of two transducers. The AE can help validate the possible complete specimen failure location identified by the DIC surface strain measurements. On the other hand, the continuous AE hits data, and DIC surface strain measurements can be useful to identify the critical timing prior to complete specimen failure. Furthermore, modal analysis using the DIC system was also adopted to study the change in specimen mode shapes, resonance frequency in the 5000 cycles of cyclic tensile loading conditions. This vibration-based evaluation can be used for periodic inspection to identify critical situations prior to catastrophic failure.

Figure 1.

The flowchart of carbon-fiber specimen complete failure prediction using surface strain measured DIC, acoustic emission, and vibration data.

Tensile loading experiment

Specimen preparation

Figure 2 shows the 0.023 in thickness of the gloss twill carbon-fiber laminate sheet that was used as the specimen material in this study. It is constructed with layers of unidirectional carbon fiber prepared in a 0°/90° orientation with a single layer of 3K 2 × 2 twill carbon fiber on each outside surface. The carbon fiber sheets use a high-quality epoxy resin that is cured at high temperatures and pressure (2021). This type of construction produces a semi-flexible but stiff laminate that has greater strength and stiffness than unidirectional carbon-fiber sheets. The material properties of the carbon-fiber laminate are shown in Table 1. Initially, the gloss twill carbon-fiber laminate was cut into test specimens according to the ASTM D3039 standard at a width of 25 mm using a band saw. In order to ensure that the complete specimen failure location occurred in the vicinity of the middle area (but not exactly at a known location) of the specimen, for each test specimen, areas of curvature were created randomly in the middle by using a belt sander, and thus each specimen has unique geometry in the area of interest (see the specimen shown in Figures 3 and 4). This unique geometry for each specimen leads to a unique stress distribution and a random failure location that needs to be predicted. The standard tensile samples do not provide a good variation for stress distribution.

Figure 2.

Carbon-fiber specimen laminate used in this work.

Table 1.

Specimen material properties (ACP Composites, 2021).

Tensile strength	266 ksi (1834 MPa)
Tensile modulus	18.3 msi (126 MPa)
Max tensile strain	10,370 (µε)
Compressive strength	193 ksi (1331 MPa)
Compressive modulus	16.5 msi (114 MPa)

Figure 3.

Static tensile strength loading experiment setup.

Figure 4.

Cyclic tensile loading experiment setup.

Static tensile strength experiment setup

A 20 kN Universal Tensile Test Machine was used in this study for three static tensile strength loading experiments, as shown in Figure 3. The DIC system used in this study was developed by Correlated Solutions Inc. For DIC measurement, a speckle pattern was applied on the surface of each carbon-fiber specimen using a speckle pattern roller. A pair of stereo cameras were set up and calibrated to cover the entire specimen surface under tensile loading. The camera frame rate was set to a maximum of 100 fps to capture the last image right before the complete specimen failure. In the meantime, a pair of piezoelectric transducers (MISTRAS R15A) was placed at each end of the specimen on the backside using hot glue. The MISTRAS USB AE Node and AEwin software were used as acoustic emission signal acquisition and post-processing equipment. The frequency range of the AE system was set at 100–600 Hz, and the amplitude threshold of the AE system was set to 45 dB, so only AE events with an amplitude of 45 dB or higher were detected in this experiment. The AE settings were selected based on the previous studies (de Groot et al., 1995; Johnson, 2000).

Cyclic tensile loading experiment

The cyclic tensile loading tests were performed at maximum loading of 1100 lbf (4900 N) force on the speckle-painted specimen for 5000 cycles. In order to understand the change in the resonance frequency and mode shapes of the specimen under cyclic tensile loading, after each 1000 tensile loading cycles, an impact hammer modal analysis was performed on the specimen (Avitabile, 2017). The mode shapes and resonance frequency of the specimen after each 1000 tensile loading cycles were extracted with the DIC system and a pair of high-speed cameras. A photograph of the cyclic tensile loading experimental setup is shown in Figure 4.

Static tensile loading test measurement

DIC measurement

Three static tensile-strength loading tests were performed on three carbon-fiber specimens with different geometrical forms in the area of interest. Prior to each static tensile strength loading test, an initial image of the specimen mounted to the mechanical grips was taken as the baseline for reference. During each of these three static tensile strengths loading tests, the DIC system measured the specimen’s surface deformation. The last image that the DIC system captured right before the complete specimen failure (the specimen broken into two pieces) was selected as the maximum specimen deformation for the surface strain analysis (see Figure 5).

Figure 5.

Specimen longitudinal (X-axis) and lateral (Y-axis) surface strain maps for the first tensile strength loading test and major AE hits locations. It should be noted that the max strain in X-axis is shown using red color in the legend, while the max strain in Y-axis is compressive and is shown using a violet color.

As can be seen in Figure 5, DIC captures all of the high longitudinal strain areas where the complete specimen failure occurred. The DIC system can simultaneously measure the longitudinal and lateral strain. However, for each of the static tensile strength loading tests, the high-strain area is within 10 mm. On the other hand, Figure 5 also shows that the high lateral strain area on the specimen surface was also captured by the DIC system, with the final breaking location falling in a high lateral strain area for strain along with the specimen. As shown in Figure 5, combining longitudinal strain maps with lateral strain maps data can provide a better tool to predict the final breaking location of the specimen. As can be seen in Figure 5, DIC cannot measure strain close to the edges of the specimen. In order to measure strain closer to the edges and to have a better spatial resolution for strain, a camera system with a higher resolution should be used. Overall, the DIC system was able to predict the failure location based on measured strain for all the specimens.

Acoustic emission measurement

One of the advantages of acoustic emission measurement is that the AE technique can measure the specimen’s subsurface defects that occurred during the tensile strength loading test by tracking the AE hits. Moreover, the AE hits can be continuously monitored so that the major subsurface carbon-fiber structure failure can be detected, which can be useful for predicting the timing of the complete specimen failure. During the several static tensile strength loading tests, the MISTRAS acoustic emission system recorded all of the AE hits with more than 45 dB amplitude and frequency between 100 and 600 Hz from the start of the tensile strength loading test to the complete specimen failure. These criteria were selected based on previously published literature (Grosse and Ohtsu, 2008). Figure 6a shows the static tensile loading test when the specimen longitudinal strain increased linearly. It can be seen in Figure 6a that there were few AE hits in the beginning, and the AE hits started to increase rapidly when the specimen was close to the complete failure point. Moreover, as shown in Figure 6b, there were some AE hits with twice the amplitude than the previous AE hits recorded during the static tensile strength test detected when the specimen was close to complete failure. Figure 6b also shows that the number and the amplitude of the AE significantly increased as the strain value approached to the breaking point. These sudden changes could be attributed to the breaking of the fibers inside the composite.

Figure 6.

(a) AE hits and longitudinal strain during first tensile strength loading test (b) Cumulative acoustic emission hits during the first tensile strength loading test.

Figure 5 also shows the failure locations predicted using AE hits and the strain map of the specimen measured using DIC. The AE source location in the longitudinal direction was estimated using the time difference of arrival between hits at the two sensors. As can be seen, the results from the two systems compare well in predicting the breaking location. The AE technique can capture AE hits with high amplitude, which can be a strong indication of major specimen subsurface failures. Also, the cumulative quantity of AE hits during the static tensile strength loading test spiked when the specimen was close to complete failure. In particular, the number of AE hits tended to spike at certain times that were close to the complete failure of the specimen, which was useful for predicting the specimen’s complete failure timing (Figure 6a). On the other hand, the specimen surface strain map can help further improve the robustness of predicting specimen complete failure timing by measuring and identifying the high strain point that is close to the specimen’s material limit prior to the complete failure. Thus, it is possible to develop a systematic approach to predict the location and timing of the specimen’s complete failure by integrating DIC and AE techniques.

Cyclic tensile loading test measurement

This section aims to study how the modal properties of a composite specimen change as it experiences cyclic loading. The cyclic loading can add some internal cracks inside the specimen, and these cracks can change its mechanical properties. The cyclic tensile loading experiment was performed at peak tensile loading of 1100 lbf force on a carbon-fiber specimen for 5000 cycles. After every 1000 cycles of tensile loading, the specimen was taken off from the tensile test machine, and one end of the specimen was fixed on the test bench. Then, the specimen was excited with a modal impact hammer, and the response of the specimen was captured using a pair of high-speed DIC cameras. The deformation data of the specimen was measured in the time-domain, and a Fast Fourier Transform (FFT) was performed on the obtained time-domain data. The specimen’s resonance frequencies and operating mode shapes were extracted by adopting the peak picking method (Avitabile, 2017).

Specimen resonance frequency measurement

From Table 2, the resonance frequencies of the specimen are generally increasing as it experiences more cyclic loading. Moreover, the incremental change for the resonance frequency of the first bending mode from 1000 to 5000 cycles is around 6%. However, the resonance frequency of the first bending mode did not change until it reached 3000 tensile loading cycles. Also, the data only shows a minimal change in the first bending mode resonance frequency (<1%) from 3000 tensile loading cycles to 5000 tensile loading cycles. Although the resonance frequency of the second bending mode is more sensitive to the tensile loading cycles, it still shows very similar results to the first bending mode. The change in the resonant frequencies can occur due to accumulated damages and material hardening (Deraemaeker and Worden, 2012; Mroz et al., 1976). While accumulated damage can reduce the resonant frequencies, the material hardening can increase them. Thus, mode shapes (displacement or strain) can be used to improve our understanding of the changes in vibration properties of the structure and to apply the vibration data for structural health monitoring.

Table 2.

The resonance frequency of the first and second bending modes during 5000 cycles of tensile loading test.

Tensile loading cycles	First mode frequency (Hz)	Second mode frequency (Hz)
1000	3.56	24.26
2000	3.56	24.35
3000	3.75	28.50
4000	3.78	28.52
5000	3.78	28.58

Mode Shape Comparison

One of the advantages of using DIC measurement technique is that 3D mode shapes of the specimen can be extracted (see Figure 7a and b). In order to plot the 1D mode shapes of the specimen, 26 points on its surface were selected (see Figure 7c). The 2D displacement of the 1^st and second bending modes (Figure 7d and e). The first two longitudinal strain modes were extracted by using DIC (Baqersad and Bharadwaj, 2018; Bharadwaj et al., 2019; Dos Santos et al., 2014). Due to the DIC camera lighting limitations, the clamping point was not captured in the mode shape extraction process. As a result, the 1^st selected point on the specimen surface is not placed on the zero-displacement point. Moreover, the 1D mode shape displacement of the first 1000 tensile loading cycles was selected as the mode shape baseline, and the rest were scaled to the same magnitude. From Figure 7a, it is clear that the first bending mode shapes of the test specimen only show a minor difference within the 5000 tensile loading cycles. On the other hand, there is a horizontal shift in the second bending mode displacement from 2000 to 3000 tensile loading cycles. In the cyclic loading, the middle section of the specimen has experienced more stress (smaller cross-section). As a result, more changes can occur in this section. That is why the second mode shape shows more changes after cyclic loading (see Figure 7e).

Figure 7.

(a) First bending 3D mode shapes of 1000 tensile loading cycles. (b) Second bending 3D mode shapes after 1000 tensile loading cycles. (c) Twenty-six selected points on the specimen surface to compare mode shapes. (d) The first bending mode in 2D0. (e) The second bending mode in 2D.

The mode shapes of a structure changes by changes in structural integrity, material properties, and boundary conditions (Baqersad et al., 2015b). A consistent boundary condition was used in this set of measurements to ensure the measured changes do not occur due to boundary conditions. On the other hand, it should also be noted that the changes in mode shapes due to the boundary conditions are different from the ones due to accumulated damages (please see Avci et al., 2021; Baqersad et al., 2015b). In order to quantify the changes in mode shapes, the modal assurance criterion (MAC) (Allemang and Brown, 1982) was used. MAC is a tool for quantifying the correlation between two vectors at all degrees of freedom. The MAC can be calculated by:

MA C_{ij} = \frac{{[{X_{i}}^{T} {X_{j}}]}^{2}}{[{X_{i}}^{T} {X_{i}}] [{X_{j}}^{T} {X_{j}}]}

(1)

where $x_{i}$ and $x_{j}$ are mode shape vectors. The MAC values close to 1.0 indicate a strong similarity between the two vectors; however, MAC values close to 0 indicate minimal or no similarity.

Table 3 shows the MAC for the first two bending displacement modes and longitudinal strain modes. The strain mode shapes using longitudinal strain values were measured using DIC (Baqersad and Bharadwaj, 2018; Bharadwaj et al., 2019; Dos Santos et al., 2014). Table 3 shows minor MAC value changes for the specimen under 5000 tensile loading cycles for the first bending mode after 3000 cycles. However, the changes in the second displacement bending mode are more apparent. On the other hand, it can be seen that the strain mode shapes are more sensitive and show more changes in the MAC values when compared to the displacement modes.

Table 3.

The MAC results comparing the specimen first and second displacement modes of the specimen after experiencing a different number of cycles to the modes at 1000 cycles.

Test cycles	MAC (displacement modes)		MAC (longitudinal strain modes)
	First mode	Second mode	First mode	Second mode
1000 cycles	1.0000	1.0000	1.0000	1.0000
2000 cycles	1.0000	0.9991	0.8685	0.8465
3000 cycles	0.9989	0.9578	0.7183	0.6863
4000 cycles	0.9988	0.9622	0.9280	0.7565
5000 cycles	0.9986	0.9516	0.8413	0.7973

Conclusion

In the field of composite material structure health inspection, non-contact methods for rapid inspection over the full structure have recently received more attention. DIC is a state-of-the-art non-contact measurement technique that is capable of identifying surface displacement and strain. On the other hand, the AE technique can be adapted to monitor the subsurface defects of the composite material. This paper studied the robustness of the DIC surface strain measurements and localization of AE hits for predicting the location of complete failure in carbon-fiber specimens. Also, this study investigated predicting the timing of the complete specimen failure based on DIC surface strain measurement and signals from AE hits in the time domain. Both DIC and AE were able to predict the location of the failure. The result showed that the combination of DIC strain and AE measurements could be used to successfully predict the timing and location of the complete specimen failure. Although a lot of improvements still need to be made to improve prediction accuracy and robustness, the DIC surface strain measurements captured the specimen complete failure locations, and AE measurements showed the potential to predict the timing of the specimen complete failure with very limited testing data. On the other hand, optical modal analysis of the specimen under cyclic loading tests using the DIC system with dual high-speed stereo cameras showed changes in the specimen’s resonance frequency and mode shapes. In particular, specimen resonance frequency and mode shapes changed with more loading cycles. The second bending mode of the specimen showed more changes when the specimen experienced more cyclic loading. The results show that DIC can simultaneously measure displacement and strain modes of the structure and can be used as a tool for periodic inspection. The results also showed that the strain mode shapes are more sensitive to damages due to fatigue in the specimen.

The results of the current work can be used for structural health monitoring of wind turbines and large composite structures. The utility-scale wind turbines can be inspected using a DIC system mounted on a drone. Retroreflective targets and patterns can be used for DIC and point tracking techniques. Furthermore, integrating acoustic emission and DIC can be used for periodic inspection of wind turbines and preventing catastrophic failures. This work will pave the way for the development of an integrated optical-acoustic measurement procedure for structural health monitoring of utility-scale wind turbines.

Footnotes

Acknowledgements

Any opinions, findings, and conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the sponsoring organizations.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research presented in this paper is partly supported by the National Science Foundation under Grant Number 1625987 (Acquisition of a 3D Digital Image Correlation System to Enhance Research and Teaching at Kettering University).

ORCID iD

Javad Baqersad

References

Allemang

Brown

(1982) A correlation coefficient for modal vector analysis. In: Proceedings of the 1st international modal analysis conference, 8–10 November 1982 Orlando: Union College Press, 110–116.

Avci

Abdeljaber

Kiranyaz

, et al. (2021) A review of vibration-based damage detection in civil structures: From traditional methods to machine learning and deep learning applications. Mechanical Systems and Signal Processing 147: 107077.

Avitabile

(2017) Modal Testing: A Practitioner’s Guide. NJ, USA: John Wiley & Sons.

Baqersad

Niezrecki

Avitabile

(2015a) Full-field dynamic strain prediction on a wind turbine using displacements of optical targets measured by stereophotogrammetry. Mechanical Systems and Signal Processing 62-63: 284–295.

Baqersad

Niezrecki

Avitabile

(2015b) Numerical and experimental analysis of the boundary conditions effects on the dynamics of wind turbines. Wind Engineering 39(4): 437–452.

Baqersad

Bharadwaj

(2018) Strain expansion-reduction approach. Mechanical Systems and Signal Processing 101: 156–167.

Baqersad

Poozesh

Niezrecki

, et al. (2017) Photogrammetry and optical methods in structural dynamics – A review. Mechanical Systems and Signal Processing 86: 17–34.

Bharadwaj

Sheidaei

Afshar

, et al. (2019) Full-field strain prediction using mode shapes measured with digital image correlation. Measurement 139: 326–333.

Brøndsted

Lilholt

Lystrup

(2005) Composite materials for wind power turbine blades. Annual Review of Materials Research 35: 505–538.

10.

Chen

Joffre

Avitabile

(2018) Underwater Dynamic Response at limited points expanded to full-field strain response. Journal of Vibration and Acoustics 140: 051016.

11.

Cigada

Mazzoleni

Tarabini

, et al. (2013) Static and Dynamic Monitoring of Bridges by Means of Vision-Based Measuring System. Topics in Dynamics of Bridges, Volume 3. New York, NY: Springer, pp.83–92.

12.

ACP Composites (2021) Carbon Fiber High Gloss Woven Sheets. Available at: https://store.acpcomposites.com/carbon-fiber-highgloss-woven-sheets?quantity=1&thickness=11.

13.

de Groot

Wijnen

PAM

Janssen

RBF

(1995) Real-time frequency determination of acoustic emission for different fracture mechanisms in carbon/epoxy composites. Composites Science and Technology 55(4): 405–412.

14.

Deraemaeker

Worden

(2012) New Trends in Vibration Based Structural Health Monitoring. New York, NY: Springer Science & Business Media.

15.

Diamanti

Soutis

(2010) Structural health monitoring techniques for aircraft composite structures. Progress in Aerospace Sciences 46: 342–352.

16.

Dos Santos

Peeters

Lau

, et al. (2014) An overview of experimental strain-based modal analysis methods. Proceedings of the International Conference on Noise and Vibration Engineering (ISMA), 15–17 september 2014, Leuven, Belgium.

17.

Eaton

Pullin

Holford

(2012) Acoustic emission source location in composite materials using delta T mapping. Composites Part A: Applied Science and Manufacturing 43(6): 856–863.

18.

Fingersh

Hand

Laxson

(2006) Wind Turbine Design Cost and Scaling Model. Golden, CO: National Renewable Energy Lab.(NREL).

19.

Goidescu

Welemane

Garnier

, et al. (2013) Damage investigation in CFRP composites using full-field measurement techniques: Combination of digital image stereo-correlation, infrared thermography and X-ray tomography. Composites Part B Engineering 48: 95–105.

20.

Grosse

Ohtsu

(2008) Acoustic Emission Testing. Springer-Verlag Belin, Heidelberg: Springer Science & Business Media.

21.

Ida

Meyendorf

(2019) Handbook of Advanced Nondestructive Evaluation. Cham, Switzerland: Springer International Publishing.

22.

Islam

Mekhilef

Saidur

(2013) Progress and recent trends of wind energy technology. Renewable and Sustainable Energy Reviews 21: 456–468.

23.

Johnson

(2000) Broad-band transient recording and characterization of acoustic emission events in composite laminates. Composites Science and Technology 60: 2803–2818.

24.

Khadka

Afshar

Zadeh

, et al. (2022) Strain monitoring of wind turbines using a semi-autonomous drone. Wind Engineering 46: 296–307.

25.

Khadka

Fick

Afshar

, et al. (2020) Non-contact vibration monitoring of rotating wind turbines using a semi-autonomous UAV. Mechanical Systems and Signal Processing 138: 106446.

26.

McCrory

Al-Jumaili

Crivelli

, et al. (2015) Damage classification in carbon fibre composites using acoustic emission: A comparison of three techniques. Composites Part B Engineering 68: 424–430.

27.

Mostafavi Yazdi

Baqersad

(2022) Mechanical modeling and characterization of human skin: A review. Journal of Biomechanics 130: 110864.

28.

Mroz

Shrivastava

Dubey

(1976) A non-linear hardening model and its application to cyclic loading. Acta Mechanica 25: 51–61.

29.

Niezrecki

Baqersad

Sabato

(2019) Digital Image Correlation Techniques for NDE and SHM. In: Ida

Meyendorf

(eds) Handbook of Advanced Nondestructive Evaluation. Cham: Springer, DOI: 10.1007/978-3-319-26553-7_47.

30.

Ozbek

Meng

Rixen

(2013) Challenges in testing and monitoring the in-operation vibration characteristics of wind turbines. Mechanical Systems and Signal Processing 41: 649–666.

31.

Panchal

Horton

Poozesh

, et al. (2019) Vibration analysis of healthy skin: Toward a noninvasive skin diagnosis methodology. Journal of Biomedical Optics 24: 015001–015011.

32.

Poozesh

Baqersad

Niezrecki

, et al. (2017) Large-area photogrammetry based testing of wind turbine blades. Mechanical Systems and Signal Processing 86: 98–115.