Measurement 164 (2020) 108026

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Nonlinear Lamb wave analysis for microdefect identification

in mechanical structural health assessment
Hanxin Chen a,b,⇑, Guangyu Zhang a, Dongliang Fan a, Lu Fang a, Lang Huang a
a
School of Mechanical and Electrical Engineering, Wuhan Institute of Technology, Wuhan 430074, China
b
Hubei Provincial Key Laboratory of Chemical Equipment, Intensification and Intrinsic Safety, China

a r t i c l e i n f o a b s t r a c t

Article history: In order to study the application of ultrasonic Lamb wave in the detection of defective plates, a nonlinear
Received 18 December 2019 Lamb wave detection system was established. The system was used to detect aluminum alloy plates with
Received in revised form 22 April 2020 different depth cracks and aluminum alloy plates with different tensile load cycles. The acquired time
Accepted 21 May 2020
domain waveforms are analyzed by Fast Fourier Transform (FFT), and the influence of two kinds of
Available online 29 May 2020
defects on the nonlinear effects of Lamb waves is obtained. The test results show that for crack defects,
the deepening of the defects will increase the ultrasonic nonlinear effect of the test piece. When the crack
Keywords:
depth exceeds 4 mm, the non-linear effect of crack defects does not increase. For fatigue crack defects, the
Nonlinear Lamb wave
Microdefect identification
occurrence of crack defects will also exacerbate the nonlinear effect of the test piece, but when the test
Mechanical structure health assessment piece is fatigue fractured, its macro defects have a smaller effect on nonlinearity.
Ultrasonic detection Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction techniques, ultrasonic testing technology uses ultrasonic waves

to encounter defects during propagation in the medium, which
Plate-like metal structures are widely used in various types of may cause interface reflection or cause changes in sound velocity
machinery and equipment, especially aluminum alloy thin plates, and energy attenuation to detect defects [5]. With the develop-
and are widely used in aerospace, high-pressure vessel and other ment of materials science and the deepening of people’s attention
equipment manufacturing industries [1]. Metal plates will cause to the early micro-damage of structures, higher requirements have
cracks, pores, Defects such as wrinkles and delaminations, under been put on the detection and evaluation methods of plate metal
the influence of the external environment during the normal work- structures that are widely used in various types of equipment.
ing process, especially under the effect of alternating loads, will The traditional linear ultrasonic inspection technology uses the lin-
gradually deteriorate the mechanical properties of the metal. In ear characteristics of sound velocity, acoustic impedance, sound
the long run, it will lead to the formation of small defects inside attenuation, and scattering to propagate flaws and performance
the metal structure, and then macro-cracks will eventually cause characteristics. However, its resolution is limited by the wave-
the structure to break. In severe cases, it may cause equipment length of the acoustic wave. It is more sensitive to volume-type
damage and bring serious catastrophic accidents [2,3]. defects that are larger than the wavelength, and it is not sensitive
In recent years, with the rapid development of science and tech- to defects such as microcracks and degradation of mechanical
nology, the quality requirements for mechanical components in properties that are smaller than its wavelength. Therefore, the
industrial production have also increased, and the demand for use of existing ultrasonic detection technology to detect micro-
the prediction of the service life of materials has also increased fatigue defects is very poor, inspired by non-linear ultrasonic
[4]. At the same time, non-destructive testing technology has also detection technology [6–8]. In recent years, nonlinear Lamb wave
been greatly developed, and the ability and accuracy of identifying monitoring technology has received increasing attention [9–14].
and locating defects is increasing. Non-destructive testing technol- Nonlinear phenomena generated by ultrasonic waves propagat-
ogy has become an extremely important key technology in indus- ing in solid materials include higher harmonics, resonance fre-
trial production. As one of several non-destructive testing quency shifts, subharmonics, and mixed frequency responses
[15].Compared with traditional ultrasonic technology, the technol-
ogy of detecting materials and structures using the propagation
⇑ Corresponding author at: School of Mechanical and Electrical Engineering, characteristics of ultrasonic guided waves in solid media is faster,
Wuhan Institute of Technology, Wuhan 430074, China. more economical and more sensitive. It has special advantages
E-mail address: pg01074075@e.ntu.edu.sg (H. Chen).

https://doi.org/10.1016/j.measurement.2020.108026
0263-2241/Ó 2020 Elsevier Ltd. All rights reserved.
2 H. Chen et al. / Measurement 164 (2020) 108026

for the early degradation of materials and the evaluation and defect detection of large-area metal plates [18]. However, the Lamb
detection of early damage inside the structure [16]. When the wave will have a dispersion phenomenon during the propagation
acoustic wave signal is transmitted in a solid material, the inherent of the sheet, and a variety of modes will be generated. The wave
nonlinearity of the material and the nonlinearity caused by the packets of these multiple modes will interfere with each other,
damage interact with it, which leads to the occurrence of ultra- making the reception and signal processing very complicated.
sonic detection acoustic wave signals Distortion, generating higher Moreover, if there are defects in the panel, the propagation of the
harmonic frequency components including second, third and Lamb wave will be more complicated, so it is obviously fundamen-
higher order harmonics [17]. Therefore, the use of ultrasonic tal to select the Lamb wave of the appropriate mode for the detec-
guided waves for defect detection is one of the important research tion of the plate material [19].
hotspots in fields of Non Destructive Testing (NDT). Lamb wave is a Deng and Pruell et al. discussed the second harmonic genera-
commonly used guided wave mode in ultrasonic nondestructive tion of nonlinear lamb waves through theoretical analysis and test
testing. Because of its advantages of small attenuation, long detec- experiments, and analyzed the feasibility of Lamb wave detection
tion distance and fast propagation speed, it is very suitable for for defect damage in the plate [20,21]. Alleyne et al. researched

Fig. 1. Dispersion curve: (a) phase velocity; (b) group velocity.

 H. Chen et al. / Measurement 164 (2020) 108026 3

the response features of lamb waves to slotted defects by using the x2 2

q2 ¼ k ð4Þ
finite element method. The reflection ratio and transmission ratio c2T
were used to evaluate the depth of the groove defects and the
The wavenumber k is numerically equal to x, where cp is the
influence of the frequency-thickness product on the propagation cp

characteristics of the lamb wave [22]. Zheng et al. The time–fre- phase velocity of Lamb and x is the circular rate. There is a simple
quency analysis of the fortune signal used in the presence or relationship between phase velocity and wavelength:cp ¼ ðx=2pÞk
absence of defects is performed by the short time Fourier trans- Eqs. (1) and (2) can be interpreted as: the relationship between
form, which effectively identifies the lamb wave modes excited the frequency x and the wave number k of the lamb wave will out-
by the plate and the mode-switching phenomenon. The results put the spectrum; or the relationship between the phase velocity
show that the abstract Lamb wave mode excited and the energy cp and the frequency x will produce a dispersion curve.
distribution center has a good correspondence with the defect The shear wave velocities of the two experimental materials
depth in the test piece [23]. Karim MR et al. Used mixed finite ele- 6061 and 5A06-O aluminum alloy used in this experiment are sim-
ment and normal mode expansion methods to study ultrasonic ilar to the phase velocities, so the same phase velocity, group
Lamb wave scattering in a plate-like crack. Liu studied the ultra- velocity dispersion curve and aluminum plate excitation angle
sonic Lamb wave propagation characteristics of thin plates, and curve are used. Aluminum alloy cL ¼ 6441m=s, cT ¼ 3224m=s. By
focused on the excitation method of ultrasonic Lamb and the vari- using matlab, the Rayleigh-Lamb frequency equation is solved by
ation of vibration displacement within the thickness of the plate. numerical method to obtain the phase velocity and group velocity
These have laid the foundation for the application of ultrasonic dispersion curve of lamb wave in aluminum alloy sheet as shown
Lamb in nondestructive testing [24]. in Fig. 1 and Fig. 2, A0–A7 is antisymmetric mode, S0–S7 is sym-
Based on the complexity of the ultrasonic Lamb wave, the signal metrical mode.
processing of the echo becomes the focus of this paper when According to the dispersion curves of the phase velocity and the
detecting the defect signal. The fast Fourier transform of the ultra- group velocity in the figure, the dispersion and multi-modal phe-
sonic Lamb wave can realize its modal recognition. This paper nomena coexist during the propagation of lamb. Observing the dis-
establishes a nonlinear ultrasonic testing system. According to persion curve, it can be seen that the frequency-thickness product
the relationship between the dispersion wave period and the wave is higher than 3 MHzmm at least three modal numbers and
velocity of aluminum alloy, the dissemination characteristics of increase the difficulty of signal analysis. When the frequency-
Lamb wave are studied. The lamb wave excitation of the appropri- thickness product is 2.5 MHzmm, the lamb mode is only A0 and
ate pattern is selected. By researching the nonlinear wave equa- S0, so it is suitable as the frequency of the excitation signal center.
tion, the relationship between the second harmonic amplitude In the light of the phase velocity dispersion bight, the excitation
and the nonlinear coefficient, the relationship between the ultra- angle bight of lamb in the aluminum alloy panel could be drawn by
sonic nonlinear coefficient and the defect depth and the fatigue 1  
Snell’s law: h ¼ sin C w =C p , as shown in Fig. 2. Where h is the
crack size are analyzed. The variation of Lamb wave with the defect excitation angle, C w is the longitudinal sound velocity of the probe,
in the aluminum alloy sheet is detected. The law of this method and C p is the phase velocity. It can be seen from the figure that the
verifies the feasibility of using this method to detect micro- incident angle of the excitation S0 and A0 mode Lamb wave are
defects in sheet metal. 50°, 75°. Because the excited A0 mode is prone to dispersion effects
in thicker plates, the A0 mode is generally used to detect thin
1.1. Lamb wave dispersion curve plates. However, due to the different operating frequencies of dif-
ferent material ultrasonic waves, the manufacturing process of the
According to the characteristics of the vibration displacement measuring probe and the material properties of the test piece are
pattern of the acoustic wave, the propagation modality of the Lamb also different, so the actual incident angle may be difference. In
wave is divided into symmetric mode and antisymmetric mode, order to ensure the accuracy of the incident angle, the transmitting
the cross-sectional displacement field of the symmetrical mode is probe and the receiving probe are kept at the same level during the
symmetrically distributed with respect to the center line, and the measurement process, and the surface coupling agent of the probe
cross-sectional displacement field of the anti-symmetric mode is in contact with the metal sheet is evenly and closely adhered.
symmetrically distributed with respect to the center line. Due to
the scattering of the guided wave and the multi-modal character-
istics, a variety of guided waves exist in one frequency segment, so 2. Selection of characterization parameters for nonlinear
it is necessary to the dispersion equation plots the dispersion curve ultrasonic testing
to find the frequency segment with the least amount of guided
wave modes. The equation describing the wave characteristics of When a certain column of ultrasonic signals propagates inside
the Lamb wave is the Rayleigh-Lamb frequency equation [25], the part to be tested, during the propagation process, the signal
and its mathematical expression is as follows:
Symmetric mode:
2
tanðqhÞ 4k pq
¼ 
2 ð1Þ
tanðphÞ 2
q2  k

Antisymmetric mode:
 2 2
tanðqhÞ q  p2
¼ ð2Þ
tanðphÞ 2
4k pq
Here p and q are given by:
x2 2
p2 ¼ k ð3Þ
c2L
Fig. 2. Aluminum plate excitation angle curve.
4 H. Chen et al. / Measurement 164 (2020) 108026

propagating interacts with the internal defects of the test piece will states. When a longitudinal wave propagates in the X direction in
eventually cause the received echo signal to be distorted. Due to a uniform solid, its one-dimensional longitudinal wave equation
the fundamental energy transfer to the harmonic waves, therefore, can be expressed as:
result in the generation of high order harmonics. Assume that in
@ 2 d 1 @T
the case of large amplitude sound waves and small deformation ¼ ð5Þ
@t 2 q @x
where d is the displacement in the x-axis direction, T(x, t) is the
stress in the x-axis direction, q is the material density, and x is
the propagation distance. Combined Eq. (4) with stress–strain non-
R  
linear relationshipT ¼ E 1 þ be þ ce2 þ    de, strain–displace-
ment relationship e ¼ @d=@x and sound velocity-density-elastic
modulus relationshipc2 ¼ E=q, the nonlinear wave equation in the
solid is obtained as:
"    3 #
2
@2d 2
2@ d 2 @ 1 @d 1 @d
 c ¼ c b þ c þ    ð6Þ
@t 2 @x2 @x 2 @x 3 @x

Fig. 3. Schematic Diagram of the Experiment Set-up.

0 10000
cycle cycle

15000 20000
cycle cycle

Fig. 4. Excitation signal time domain waveform. Fig. 6. Fatigue tensile crack.

1mm 2mm 3mm

4mm 5mm

Fig. 5. Artificial crack.

 H. Chen et al. / Measurement 164 (2020) 108026 5

The right portion of formula (6) to retain the items available c: The solution of the nonlinear acoustic wave equation in the
2 2  2 2
2 solid obtained by the progressive analysis method is:
@ d @ d @d @ d @d @ d
 c2 2 ¼ c2 b þ c2 c ð7Þ 2
@t 2 @x @x @x 2 @x @x2 d ¼ A0 sinðkx  xt Þ  8b xk A20 cos2ðkx  xt Þ
c 2
ð11Þ
Set its solution to: þ 24 xk A30 ½cos3ðkx  xt Þ þ 3cosðkx  xt Þ

dðx; t Þ ¼ dð0Þ þ bdð1Þ þ cdð2Þ ð8Þ In the Eq. (10), the fundamental amplitude A1 , the second har-
monic amplitude A2 and the third harmonic amplitude A3 are:
Bringing the formula (8) into the Eq. (7) gives: 8

2 ð1Þ 
2 ð2Þ  > A ¼ A0
2 ð0Þ
@ d 2 ð0Þ 2 ð1Þ 2 ð2Þ < 1
@t 2
 c2 @ @xd2þ b @ @td2  c2 @ @xd2 þ c @ @td2  c2 @@xd2 2
A2 ¼ 8b xk A20 ð12Þ

ð0Þ 2 2 ð0Þ ð9Þ >
: c 3
ð0Þ 2 ð0Þ
¼ c2 b @d@x @@xd2 þ c2 c @d@x @ d A3 ¼ 24 xk A30
@x2
From Eq. (11), the relationship between the second-order non-
According to the formula (8):
linear coefficient b and the third-order nonlinear coefficient c can
8
2 ð1Þ 
> 2 ð1Þ ð0Þ
< b @ @td2  c2 @@xd2  c2 b @d@x @ @xd2
2 ð0Þ be derived:
8


2 ð10Þ < b ¼ k8A 2
>
: c @2 d2ð2Þ  c2 @2 dð22Þ  c2 c @dð0Þ @ 2 dð20Þ 2 2
A x
@t @x @x @x
1
ð13Þ
:c ¼ 24A3
k3 A31 x

Fig. 7. Fundamental signal of artificial crack test: (a) time domain signal (b) frequency domain signal.
6 H. Chen et al. / Measurement 164 (2020) 108026

where k ¼ 2pf =c is the wave number and x is the sound wave prop- @2u @r
q ¼ ð15Þ
agation distance. @t 2 @x
The main purpose of this paper is to establish the correspon-
where q is the density of the material, r is the stress, and x is the
dence between crack depth and ultrasonic nonlinearity of alu-
propagation distance. The relationship between it and the strain e
minum alloy surface. Because the nonlinear effect of the
nonlinear Lamb wave is extremely insignificant in the generation is eðx; t Þ ¼ @u@x
ðx;t Þ

of the third harmonic phenomenon. The following focuses on the The nonlinear wave equation of uðx; tÞ can be obtained from Eqs.
second-order nonlinear coefficients. (5) and (6)[26]:
According to Hooke’s law, the stress r has a nonlinear relation- @2u @2u @u @ 2 u
ship with the strain e, and its one-dimensional equation is as q ¼ E 2 þ 2Eb ð16Þ
@t 2 @x @x @x2
follows:
The stress r value of the formula (5) is generally considered to
r ¼ Eeð1 þ be þ   Þ ð14Þ
depend on the second-order constant. Solve using the singular per-
where E is the Young’s modulus and b is the second-order nonlinear turbation method. Suppose the displacement u is determined by:
coefficient. u ¼ u0 þ u0 ð17Þ
Without considering attenuation, the one-dimensional wave
equation is:

Fig. 8. Artificial crack test second harmonic signal: (a) time domain signal (b) frequency domain signal.
 H. Chen et al. / Measurement 164 (2020) 108026 7

where u0 is the initial displacement of the excitation wave and When the material properties, wavenumber k and propagation
u0 is the first-order perturbation solution. Suppose u0 is a sine wave distance x are determined, b is proportional to AA22 , introducing a rel-
of a single frequency: 1

ative nonlinear coefficient

u0 ¼ A1 cosðkx  xtÞ ð18Þ
A2
where k is the wave number and the second-order perturbation
b0 ¼ ð23Þ
A21
solution is obtained as follows:
The nonlinear source of the sample to be tested consists of two
u ¼ u0 þ u0 ¼ A1 cosðkx  xt Þ  A2 sin2ðkx  xt Þ ð19Þ
parts: one is the inherent nonlinearity of the metal material, and
Among them the other is the acoustic nonlinearity caused by the near surface
defects of the metal, so the relationship between the nonlinear
b 2 2
A2 ¼ A k x ð20Þ coefficient and the depth of the defect can be established.
8 1
Thereby
3. Test equipment and results analysis
A2
b¼8 ð21Þ
2
k A21 x 3.1. Test device

For the nonlinear coefficient of lamb, it is necessary to introduce The detection principle of the nonlinear system detection is
the scale transform coefficient c, so the lamb wave nonlinear coef- shown in Fig. 3. It consists of a RAM-5000-SNAP nonlinear high-
ficient is [27]: energy ultrasonic test system, a PC, a 50 X load, an attenuator, a
A2 high-pass and low-pass filter, and a preamplifier. The nonlinear
b¼8 2
c ð22Þ system is connected to the PC, and the parameters are controlled
k A21 x
by software on the computer, and the mode of one-shot and

Fig. 9. Fundamental signal of fatigue crack test: (a) time domain signal (b) frequency domain signal.
8 H. Chen et al. / Measurement 164 (2020) 108026

two-receipt is adopted. The signal generated by the ultrasonic non- curve of group velocity and phase velocity, this experiment selects
linear detection system is emitted by the transmitting end. After a 2.5 MHz sinusoidal pulse train as the center frequency of the
the impedance and the low-pass filter, the ultrasonic signal excited excitation signal, and uses the Hanning window to adjust the exci-
by the transmitting transducer is transmitted to the receiving tation signal. The energy of the excitation signal after windowing
transducer through the test piece, and the signal wave propagates can be more concentrated, the flap is significantly inhibited and
in the metal plate. Waveform distortion is generated, resulting in reduces energy leakage, resulting in better signal reception. In
secondary harmonics and above. Because the higher harmonic sig- order to ensure that the frequency component of the response sig-
nal is weak, it is easily covered by the noise attached to the system nal is purer, the number of bursts should be as much as possible,
and the fundamental signal. Therefore, the fundamental wave and and then the received signals cannot be overlapped, so 15 cycles
the second harmonic signal are respectively received, where the of excitation signals are selected. Therefore, the excitation signal
fundamental wave signal is directly returned to the receiving 1 finally selected in this experiment is 15 cycles single-frequency
of the system without being processed, and the second harmonic sinusoidal pulse train modulated by Hanning window, and Fig. 4
signal is returned to the receiving 2 via the high-pass filter and is the time domain waveform of the excitation signal.
the signal amplifier. The oscilloscope can simultaneously display
acquire excitation and receive signals [28,29]. 3.2. Test piece
This experiment uses the same side ultrasonic excitation receiv-
ing method. The ultrasonic signal excited by the excitation probe The test piece materials are 6061 aluminum alloy and 5A06-O
propagates along the direction of the test piece and is received aluminum alloy, and the defects are divided into two types: (1)
after receiving the probe. An angle-adjustable probe is accustomed 6061 aluminum alloy, the size is 250 mm  100 mm  10 mm,
to impel the fixed mode lamb wave. Referring to the dispersion and the method of wire cutting in the width direction in the middle

Fig. 10. Fatigue crack test second harmonic signal: (a) time domain signal (b) frequency domain signal.
 H. Chen et al. / Measurement 164 (2020) 108026 9

position of the test piece respectively Machining depths of 0 mm,

1 mm, 2 mm, 3 mm, 4 mm, 5 mm, width 0.1 mm, excited S0
mode, to study the relationship between nonlinear coefficient
and crack depth; (2) 5A06-O aluminum alloy, the size is
300 mm  50 mm  2 mm, a triangular notch is opened in the mid-
dle of the test piece, make stress concentration easy to stretch. And
the fatigue tensile test is carried out. The vibration period is 0,
10,000, 15,000, 20,000 cycles. After the test piece was shaken
18,618 weeks, the test piece broke in the working area to form a
macro crack, excited A0 mode. Observe the effects of different fati-
gue cycles on nonlinear effects. Figs. 5 and 6 is a picture of two
defects under a PVA ultrasound microscope.

3.3. Analysis of experimental results

The nonlinear Lamb wave detection platform used in this exper-

iment can realize the reception of the fundamental and second har-
monic time domain waveforms, the excitation transducer center
frequency is 2.25 MHz, receive transducer center frequency is
5 MHz. During the experiment, the measurement conditions of
the fundamental wave and the second harmonic amplitude should
be consistent. For the test pieces with different depth cracks and 10 Fig. 11. Regularization nonlinear coefficient and crack depth relationship.
test pieces with different fatigue degrees, the fundamental wave
and the second harmonic are measured in sequence. The time
domain waveform is fast Fourier transformed, the biggest advan- relative coefficient of the unprocessed test piece is b00 . The relative
tage of this algorithm is: not only can you see which frequency nonlinear coefficient parameters of the wire-cut test piece are reg-
components are included in the signal, but also you can observe b0
ularized, that is b00
. As shown in Tables 1 and 2.
the occurrence time of the frequency components, and its calcula-
The correspondence between the regularization relative nonlin-
tion is small, and the transformed fundamental amplitude A1 and
ear coefficient and the defect size of the defective test piece
second harmonic amplitude A2 are shown in Figs. 7, 8, 9, 10.
obtained by the above process is shown in Figs. 11 and 12.
The fundamental and second harmonic amplitudes of artificial
It can be seen from the Fig. 11 that this curve can be divided
cracks and fatigue cracks are shown in figure. It can be seen that
into two stages: the first stage before 4 mm, the regularization rel-
the amplitude of the fundamental wave does not change much,
ative nonlinear increases with the increase of the crack depth; the
and the amplitude of the second harmonic is sensitive to the size
second stage: after the depth of 4 mm, the ultrasonic nonlinear
of the defect damage. So the second harmonic amplitude can
coefficient appears the fluctuations and shows downward trend,
roughly characterize the depth of the defect, but the theoretical
indicating that the change in the relative nonlinear coefficient does
basis of the previous nonlinear ultrasonic detection shows that
not increase linearly as the depth of the defect increases. This is
the relative ultrasonic nonlinear coefficient is usually used to char-
because the ultrasonic lamb wave is a plane strain wave generated
acterize the nonlinear effect, because this is the performance of the
in the free plate, which is generated at the same order of magni-
comprehensive performance of the fundamental amplitude and
tude as the thickness of the medium and the excitation wave-
the second harmonic amplitude, and better reflect the variation
length, so the ultrasonic lamb wave is sensitive to the structure
of the nonlinear effect. Bringing the measured value into equation
of the detecting thin plate. Therefore, when the metal plate
(23) yields to get relative nonlinear coefficient b0 . The nonlinear

Table 1
Regularization relative nonlinear coefficient value.

Crack depth /mm 0 1 2 3 4 5

A1 /V 1:49  10 3
1:50  10 3
1:60  10 3
1:54  10 3
1:55  103
1:60  103
A2 /V 5:9  104 8:5  104 1  103 1:29  103 1:47  103 1:35  103
0
b 1 1.664 1.724 2.402 2.700 2.327
0
b0

Table 2
Regularization relative nonlinear coefficient value.

Fatigue cycle/10,000 Times 0 1 1.5 2

A1 /V 3:59  103
3:92  10 3
3:83  10 3
3:11  103
A2 /V 1:42  104 5:01  104 5:78  104 1:24  104
b 1 2.963 3.578 1.183
b0
10 H. Chen et al. / Measurement 164 (2020) 108026

Data curation, Investigation. Lu Fang: Data curation, Investigation.

Lang Huang: Data curation, Investigation.

Declaration of Competing Interest

The authors declare that they have no known competing finan-

cial interests or personal relationships that could have appeared
to influence the work reported in this paper.

Acknowledgement

This work was supported by the Special Major Project of the

Ministry of Science and Technology of Hubei Province of China
(Grant No. 2016AAA056), Major project of Hubei Provincial Depart-
ment of Education (Z20101501) and the National Natural Science
Foundation of China (Grant 51775390).

References

[1] Hanxin Chen, Yunfei Shang, Kui Sun, Multiple fault condition recognition of
gearbox with sequential hypothesis test, Mech. Syst. Signal Process. 40 (2)
(2013) 469–482.
[2] Liu Yang, Hanxin Chen, Fault diagnosis of gearbox based on RBF-PF and particle
Fig. 12. Regularization nonlinear coefficient and fatigue cycle relationship. swarm optimization wavelet neural network, Neural Comput. Appl. 31 (9)
(2019) 4463–4478.
[3] Hanxin Chen, Lang Huang, Liu Yang, Yongting Chen, Jinmin Huang, Model-
exceeds a certain thickness, the propagation of the lamb wave is based method with nonlinear ultrasonic system identification for mechanical
limited. structural health assessment, Trans. Emerg. Telecommun. Technol. 3955
From the Fig. 12, the variation trend of the Lamb wave nonlin- (2020). http://doi.org/10.1002/ett.3955.
[4] Hanxin Chen, Yongting Chen, Liu Yang, Intelligent early structural health
ear coefficient in the whole crack propagation process of the spec- prognosis with nonlinear system identification for RFID signal analysis,
imen, the relative nonlinear coefficient of the Lamb wave increases Comput. Commun. (2020), https://doi.org/10.1016/j.comcom.2020.04.026.
first and then decreases with the increase of the fatigue degree of [5] Li Zeng, Geng Chen, Hanxin Chen, Comparative study on flow-accelerated
corrosion and erosion-corrosion at a 900 carbon steel bend, Materials,
the specimen. The first three tests The number of parts increases in
Materials 13 (7) (2020), In this issue 10.3390/ma13071780.
turn, and it shows a downward trend after the occurrence of [6] Zhou Zhengqian, Liu Siming, Research, application and development of
macroscopic cracks in the fourth block, indicating that the nonlin- nonlinear nondestructive testing technology, J. Mech. Eng. 47 (8) (2011) 2–11.
[7] Guo Yi, Shi Xiaohong, Wang Guanglong, Nonlinear ultrasonic testing of micro-
ear ultrasonic effect is sensitive to micro defects and does not have
nano-scale cracks, Edit. Off. Opt. Precis. Eng. 19 (1) (2011) 132–137.
an advantage for the measurement of macroscopic cracks. [8] Bingsheng Yan, Wu. Bin, Cunfu He, et al., Early fatigue damage of magnesium
alloy on-line monitoring by nonlinear ultrasonic, Adv. Mater. Res. 139–141
(2010) 194–197.
4. Conclusion [9] Deng Mingxi, Xiang Yanxun, Liu Liangbin, Time-domain analysis and
experimental examination of cumulative second-harmonic generation by
primary Lamb wave propagation, J. Appl. Phys. 109 (2011).
(1). The FFT is used to analyze and convert the fundamental and [10] Deng Mingxi, A nonlinear ultrasonic lamb wave method for evaluating surface
second harmonic signals collected by the two kinds of properties of solid sheets, Appl. Acoust. 32 (4) (2013) 262–269.
defects, which can effectively observe the amplitude varia- [11] M.F. Müller, J.Y. Kim, Q.U. Jianmin, et al., Characteristics of second harmonic
generation of Lamb waves in nonlinear elastic plates, J. Acoust. Soc. Am. 127
tion of the received signal with the defect of the test piece.
(4) (2010) 2141–2152.
(2). By analyzing the guided wave received signals of different [12] D. Dutta, H. Sohn, K.A. Harries, et al., A nonlinear acoustic technique for crack
depth defects, the defect depth has a good correspondence detection in metallic structures, Struct. Health Monit. 8 (3) (2009) 251–262.
[13] N. Matsuda, S. Biwa, Phase and group velocity matching for cumulative
with the amplitude of the guided wave at a certain distance.
harmonic generation in Lamb waves, J. Appl. Phys. 109 (2011) 094903.
The relative nonlinear coefficient of the Lamb surface wave [14] K.H. Matlack, J.Y. Kim, L.J. Jacobs, et al., Experimental characterization of
can well represent the size of the depth of the defect. But efficient second harmonic generation of Lamb wave modes in a nonlinear
the relative nonlinear coefficient parameter does not elastic isotropic plate, J. Appl. Phys. 109 (2011) 014905.
[15] K. Jhang, Nonlinear ultrasonic techniques for nondestructive assessment of
increase linearly with the increase of crack depth, so this micro damage in material: a review, Int. J. Precis. Eng. Manuf. 10 (1) (2009)
method can only be applied to the near surface detection 123–135.
of sheet metal. [16] Chen Hanxin, Lu Yanjun, Tu Ling, Fault identification of gearbox degradation
with optimized wavelet neural network, Shock Vib. 20 (2) (2013) 247–262.
(3). During the intensification of fatigue cracks in thin metal [17] Hanxin Chen, Wenjian Huang, Jinmin Huang, Chenghao Cao, Liu Yang, Yibin He
plates, the non-linear effects of cracks increase with the and Li Zeng, Multi-fault condition monitoring of slurry pump with principle
deepening of fatigue, but they show a downward trend after component analysis and sequential hypothesis test, Int. J. Pattern Recogn.
Artif. Intell., https://doi.org/10.1142/S0218001420590193.
fatigue fracture of the test piece. The macro-crack at the [18] Chen Jun, Li Zhihao, Lin Li, Experimental study on lamb wave detection in
time has little effect on the nonlinear effect. aluminum plate, Appl. Acoust. 30 (2) (2011) 98–104.
[19] Hanxin Chen, Dong Liang Fan, Lu Fang, Wenjian Huang, Jinmin Huang,
Chenghao Cao, Liu Yang, Yibin He and Li Zeng, Particle swarm optimization
CRediT authorship contribution statement algorithm with mutation operator for particle filter noise reduction in
mechanical fault diagnosis, Int. J. Pattern Recogn. Artif. Intell. https://doi.org/
10.1142/S0218001420580124.
Hanxin Chen: Conceptualization, Data curation, Formal analy-
[20] M. Deng, Analysis of second-harmonic generation of Lamb waves propagating
sis, Funding acquisition, Investigation, Methodology, Project in layered planar structures with imperfect interfaces, Appl. Phys. Lett. 88 (22)
administration, Resources, Software, Supervision, Validation, (2006) 221902.
Visualization, Writing - original draft, Writing - review & editing. [21] C. Pruell, J.Y. Kim, J. Qu, et al., Evaluation of fatigue damage using nonlinear
guided waves, Smart Mater. Struct. 18 (3) (2009) 035003.
Guangyu Zhang: Data curation, Investigation, Methodology, Writ- [22] D.N. Alleyne, P. Cawley, The interaction of Lamb waves with defects, IEEE
ing - original draft, Writing - review & editing. Dongliang Fan: Trans. Ultrason. Ferroelectr. Freq. Control 39 (3) (1992) 381–397.
 H. Chen et al. / Measurement 164 (2020) 108026 11

[23] Zheng Xiangming, Yang Qi, Hao Guofa, Lamb wave detection of seam defects in Dongliang Fan graduated as B.S in 2016 and is master
thin metal sheets, Non-Destruct. Test. 29 (11) (2007) 630–649. student in School of Mechatronics Engineering at
[24] Hanxin Chen, Dongliang Fan, Jinmin Huang, Wenjian Huang, Guangyu Zhang, Wuhan Institute of Technology in China. His research
Lang Huang, Finite element analysis model on ultrasonic phased array interests are Structural Health Monitoring, Finite
technique for material defect time of flight diffraction detection, Sci. Adv. element analysis for ultrasonic detection etc.
Mater. 12 (5) (2020) 665–675.
[25] J.L. Rose, He Cunfu, Wu Bin, Wang Xiuyan translation. Ultrasonic in solids.
Science Press, Beijing, 2004, pp. 89–91.
[26] M.A. Breazeale, J. Philip, Determination of third-order elastic constants from
ultrasonic harmonic generation measurements, Phys. Acoust.: Principles
Methods 17 (1984) 1–60.
[27] Chao Zhou, Ming Hong, Zhong Qingsu, Evaluation of fatigue cracks using
nonlinearities of acousto-ultrasonic waves acquired by an active sensor
network, Smart Mater. Struct. 22 (2013) 015018.
[28] Liu Yang, Hanxin Chen, et al., A novel time-frequency-space method with
parallel factor theory for big data analysis in condition monitoring of complex
system, International Journal of Advanced Robotic Systems 17 (2) (2020), Lu Fang graduated from Yangtze Normal University
https://doi.org/10.1177/1729881420916948. Normal University in 2017 and is master student in
[29] Yibing He, Wen-Hao Xiong, Han-Xin Chen, Cheng-Hao Cao, Wengjian Huang, School of Mechanical and Electrical Engineering at
Liu Yang, Zeng Li, Qiao-Shen Dai, Yu-Chen Chen, Image quality enhanced Wuhan University of technology in China. His research
recognition of laser cavity based on improved random hough transform, J. Vis. interests are Structural Health Monitoring, fault
Commun. Image Represent. (2020). https://doi.org/10.1016/j.jvcir.2019. diagnosis, and non-destructive testing etc.
102679.

Hanxin Chen obtained PHD in School of Mechanical and

Aerospace Engineering at Nanyang Technological
University in Singapore from July, 2001 to February,
2006. He was post-doc research fellow in Department of
mechanical engineering at University of Alberta in
Canada from March, 2006 to March, 2008. He was
research associate in School of Physics at University of Lang Huang graduated as B.S in 2017 and is master
Windsor in Canada in 2012 and research scientist in student in School of Mechatronics Engineering at
Department of Control at Sheffield of University from Wuhan Institute of Technology in China. His research
December, 2012 to May, 2015. He is distinguished interests are Structural Health Monitoring, Finite
professor as ‘‘Chutian Scholar” of Hubei Province at element analysis for ultrasonic detection etc.
Wuhan Institute of Technology from April, 2008.
His research area is about condition monitoring and fault diagnosis of engineering
system, structural health monitoring (SHM) etc.

Guangyu Zhang graduated as B.S in 2016 and is master

student in School of Mechatronics Engineering at
Wuhan Institute of Technology in China. His research
interests are Structural Health Monitoring, Finite
element analysis for ultrasonic detection etc.