Available online at www.sciencedirect.

com

ScienceDirect
Physics Procedia 70 (2015) 488 – 491

2015 International Congress on Ultrasonics, 2015 ICU Metz

Enhancement of phased array ultrasonic signal in composite

materials using TMST algorithm
BENAMMAR Abdessalem*, DRAI Redouane, KECHIDA Ahmed, DRIS Lyamine,
CHIBANE Farid
Welding and NDT Research Center (CSC), BP 64, Cheraga, 16014 Alger, Algeria

Abstract

In this paper, we apply a new technique for the ultrasonic phased array signal enhancement. It is based on the threshold modified
S-transform (TMST). The signal processing algorithms generally give very satisfactory results on synthetic signals verifying the
implicit or explicit hypotheses on which they are constructed. The obtained performances on the real signals can be however
different radically. Time–frequency analysis methods are mainly used to improve the defects detection resolution. Significant
performance enhancement is confirmed when the proposed approach is tested with the simulation of the B-scan signals contain a
closer delamination to the front face. The experimental results show that the TMST Algorithm can enhance the quality of image
provided by composite materials contained delamination defect.
© 2015
© 2015TheTheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B.V. B.V.
This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the Scientific Committee of 2015 ICU Metz.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Scientific Committee of ICU 2015
Keywords: Ultrasonic Phased Array, Composite Materials, S-transform, TMST Algorithm;

1. Introduction

Fiber reinforced laminated Composite materials are increasingly used in aerospace, naval and automotive
applications due to their high specific strength, high stiffness, and high strength for low weight. However, one
weakness of such materials is their sensitivity to impacts, which may cause severe and harmful damages (Castaings
and al., 2012; Benammar and Drai, 2014). With the increased usage of fiber reinforced materials in structural

* Corresponding author. Tel.: +21321342019; fax: +21321342019.

E-mail address: a.benammar@csc.dz abs_benammar@yahoo.fr

1875-3892 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Scientific Committee of ICU 2015
doi:10.1016/j.phpro.2015.08.292
 Benammar Abdessalem et al. / Physics Procedia 70 (2015) 488 – 491 489

materials, it has become all the more important for the NDT community to develop viable technique which can be
used to test these materials while they are in service. Typical NDT techniques which have been used to try and
reconstruct signals and/or images of defects inside the metals and composite materials are visual optical inspection,
radiography, magnet particle inspection, eddy current testing, ultrasonic testing, acoustic emission testing,
microwave testing, etc (Edwards and al., 2006). Ultrasonic phased array system for non-destructive testing (NDT)
has been used increasingly in recent years. Compared with conventional single element transducers, the ultrasonic
phased array can perform multiple inspections without the need of reconfiguration, which obtains better sensitivity
and coverage (Xiao and al., 2012). Many techniques were presented by various researchers for detect, analyze or
classify the ultrasonic signals and/or ultrasonic images. One of the most widely used methods in signal processing is
time frequency analysis, such as STFT, Wigner-Ville distribution (Rodriguez and al., 2004) and WT (Song and Que,
2006) etc. have been used to deal with this situation by examining the signal in joint time frequency domain.
However, this paper focuses on time frequency analysis techniques to enhance the B-scan signals. The STFT most
often used but it cannot track the signal dynamics properly for non-stationary signal due to limitation of fixed
window width. The limitations of the STFT have motivated the development of alternative techniques, such as S-
transform. The S-transform is an invertible time frequency spectral localization technique that combines elements of
wavelet transforms and short-time Fourier transform. the advantage of S-transform provides multiresolution analysis
while retaining the absolute phase of each frequency (Huda and al., 2013). The time frequency distribution of
ultrasonic echoes suffers from three major problems: measurement noise, overlapping echoes and changes in signal
characteristics (Benammar and al., 2014). The S-transform with modified Gaussian window or Modified S-transform
(MST) is essentially a variable window STFT whose window width varies inversely with the frequency. The MST
produces a time-frequency representation of a time varying signal by uniquely combining the frequency dependent
resolution with simultaneously localizing the real and imaginary spectra. Since the MST provides the local spectrum
of a signal the time averaging of the local spectrum gives the Fourier spectrum. In this work, our contributions are to
apply a new algorithm, called threshold modified S-transform (TMST) in order to enhance the b-scan images. The
rest of this paper is organized as follows. Section II gives the threshold modified S-transform (TMST) algorithm.
Section 3 shows the results from simulated and real data are presented to illustrate the method. Finally, conclusions
are given in section 4.

2. Threshold modified S-transform algorithm

2.1. Modified S-transform

In the modified S-transform, we are using a different scaling rule for the Gaussian window. This can be achieved
by getting the scaling parameter Ȗ defined above to vary linearly with the frequency to get better progressive control
of the window width as follows: ߛሺ݂ሻ ൌ ݂ܽ ൅ ܾ (Stockwell, 2007)
Where a is the slope and b is the intercept for a linear change in frequency. Hence, the resolution in time and in
frequency depends on both the parameters a and b. These parameters are able to control the Gaussian window
length. With this new formulation the modified S-transform becomes:
ஶ
ܵሺ߬ǡ ݂ǡ ܽǡ ܾሻ ൌ ‫ି׬‬ஶ ‫ݔ‬ሺ‫ݐ‬ሻ‫ݓ‬ሺ߬ െ ‫ݐ‬ǡ ݂ǡ ܽǡ ܾሻ݁ ି௜ଶగ௙௧ ݀‫ ݐ‬. Where w denotes the window function of the modified S-
೑మ ሺഓష೟ሻమ
ȁ௙ȁ ൤ି ൨
transform expressed by: ‫ݓ‬ሺ߬ െ ‫ݐ‬ǡ ݂ǡ ܽǡ ܾሻ ൌ ݁ మሺೌ೑శ್ሻమ .
ξଶగሺ௔௙ା௕ሻ
The computation of the MST is efficiently implemented using the convolution theorem and Fourier transform.
The following steps are used for the computation of Modified S-transform:

1.Compute the DFT of the signal x(k) using FFT software routine and shift spectrum X[m] to X[m+n].
2.Compute the Gaussian window function of the required frequency n.
3.Compute the inverse Fourier transform of the product of DFT and Gaussian window function to give the MST
matrix.
The output of the MST is an n*m matrix. The rows pertain to frequency and columns indicate time. Each column
thus represents the “local spectrum” for that point in time. From the MST matrix we obtain the frequency-time
490 Benammar Abdessalem et al. / Physics Procedia 70 (2015) 488 – 491

contours having the same amplitude spectrum and these contours can be used to visually classify the nature of the
signal and its change of frequency and the harmonic content if any during the distortion.

2.2. The Proposed TMST algorithm

The evaluation of arrival time and center frequency of multiple echoes in an ultrasonic signal is dependent on the
time frequency resolution of the S-transform of the signal. These two parameters are very important in ultrasonic
detection. MST is highly suitable in separating the noise elements from the signal components as it acts as a power
discriminator (Biswal and al., 2011). Since the power content of the noise is usually lower than ultrasonic signal, it
is possible to remove them using MST. But for a better separation of close echoes, we need to develop a new
algorithm based on the modified S-transform and a thresholding technique. This algorithm is called threshold
modified S-transform (TMST). The iterative TMST algorithm that we propose consists of several stages. The first
stage of the algorithm is to compute the MST of the received signal x(t). In the second stage, we localize Maximum
of XMST (the most dominant echo in the MST). Then, we isolate it by applying a window filter in the temporal
field located in the dominant echo position. The isolation window width is equivalent to the system impulse
response. This process is repeated until all the remaining echoes are detected. The threshold value in TMST
algorithm is the average of signal expressed by: Š”‡•Š‘Ž† ൌ σே ௧ୀଵȁ‫ݔ‬ሺ‫ݐ‬ሻȁΤܰ , where N is the length of the signal.
The number of iterations in the TMST depends on the number of echoes. The computational time in the TMST
algorithm is almost similar with other time-frequency methods.

3. Results and discussions

In this section, the results of the Threshold modified S-transform algorithm applied to ultrasonic B-scan signal
are discussed.

3.1. Simulation study

In order to apply the proposed algorithms, a numerical experiment is set up simulating a material with three
defects. The first defect is very close in time to the front face, second defect is in the middle of the material, and the
third defect is close to the back wall face as illustrated by the B-scan image in Fig. 1a.

B-scan STFT
A-scan STFT

5 5
1 1
10 10
0.8 0.8
Am plitude

Amplitude

15 0.6 15
0.6
20 0.4 20
0.4
25 0.2 25
0.2
30 0 30 0
0 5 10 15 20 0 5 10 15 20
200 400 600 800 1000 200 400 600 800 1000
Time in microseconds Time in microseconds
ST
ST TMST TMST
5
1 5 1
10
0.8 10 0.8
Amplitude

A m plitude

15
0.6 15 0.6
20 0.4 20 0.4
25 0.2
25 0.2
30 0
0 5 10 15 20 30 0
200 400 600 800 1000 0 5 10 15 20
Time in microseconds 200 400 600 800 1000 Time in microseconds

Fig. 1. (a) Original B-scan with a SNR of 5dB; (b) A scan number 16; (c) STFT of the image B-scan in (a); (d) STFT amplitude of the
signal A scan number 16; (e) S-transform of the image B-scan in (a); (f) S-transform amplitude of the signal A scan number 16; (g) TMST
algorithm of the image B-scan in (a) (threshold=0.18); (h) TMST algorithm amplitude of the signal A scan number 16.

Figure (2.b) shows that despite a high rate of SNR (5dB) the proposed TMST algorithm can significantly
improve the image which gives a good resolution and ensures proper detection.
 Benammar Abdessalem et al. / Physics Procedia 70 (2015) 488 – 491 491

3.2. Experimental Data Analysis

The ultrasonic data was obtained from a Carbon-fiber-reinforced polymer composite material (CFRP) with
delamination defect close to the front face. The part is provided by an aircraft manufacturer company which is
achieved by sticking the unidirectional layers with epoxy one layer on the other altering the orientation from (0°,
45°, 0°). The thickness in the undamaged zone is 2.67 mm and 3.3 mm in the damaged zone. The ultrasonic signals
were obtained by OMNISCAN MX acquisition system by Olympus Company, the A-scan signals which were
considered as original signals exported by TOMOVIEW and represent as B-scan image. The used transducer is
composed of 32 linear piezoelectric elements with centered frequency of 5Mhz. Longitudinal waves are used and we
recall that the sound velocity in this material is Vsample = 2830 m/s.
B-scan STFT ST TMST

5 5 5 5
10 10 10 10
15 15 15 15
20 20 20 20
25 25 25 25
30 30 30 30

100 200 300 400 500 600 200 400 600 800 1000 100 200 300 400 500 600 100 200 300 400 500 600

Fig. 2. (a) Experimental B-scan image obtained from CFRP composite material (delamination close to the front face); (b) STFT of the image B-
scan in (a); (c) S-transform of the image B-scan in (a); (d) TMST algorithm of the image B-scan in (a) (threshold=0.18).

This example shows that only the TMST algorithm provides good improvement in image B scan which gives a
good detection of delamination defect.

4. Conclusions

In this paper, we have applied a new technique for the ultrasonic phased array signal and compared with other
methods for improved ultrasonic flaw detection using the frequency distributions. The approach to process signals
was to introduce the threshold in the MST. The obtained results in experimental signals confirm that the algorithm
gives a proper detection of delamination in the composite material. The analytical and experimental studies verify
the feasibility of the time frequency techniques for NDE applications.

References

Michel Castaings, Dilbag Singh, Philippe Viot, “Sizing of impact damages in composite materials using ultrasonic guided waves” NDT&E
International, 46 (2012) 22–31.
A.Benammar and R. Drai. Ultrasonic Flaw Detection in Composite Materials Using SSP-MPSD Algorithm, JEET Vol. 9, No.5: 1753-1761, 2014.
R.S. Edwards, S. Dixon, X. Jian, “Characterisation of defects in the railhead using ultrasonic surface waves”,NDT&E Int. 2006 (39) 468-475.
Kun Xiao, Qiang Wang, Dong Hu, “Post Signal Processing of Ultrasonic Phased Array Inspection Data for Non-destructive Testing”, Procedia
Engineering 43 ( 2012 ) 419–424.
M.A. Rodriguez, J.L. San Emeterio, J.C. Lazaro, A. Ramos, Ultrasonic flaw detection in NDE of highly scattering materials using wavelet and
Wigner–Ville transform processing, Ultrasonics 42 (1–9) (2004) 847–851.
S. Song, P. Que, Wavelet based noise suppression technique and its application to ultrasonic flaw detection, Ultrasonics 44 (2) (2006) 188–193.
N.H.T.Huda, A.R. Abdullah, M.H Jopri, “Power Quality Signals Detection Using S-Transform”, IEEE 2013, 978-1-4673-5074-7/13.
A.Benammar, R. Drai, A. Guessoum, Ultrasonic flaw detection using threshold modified S-transform, Ultrasonics 54 (2014), pp676-683.
R.G. Stockwell, L. Mansinha, R.P. Lowe, Localisation of the complex spectrum: the S transform, IEEE Signal Process. 44 (1996) 998–1001.
S. Assous, B. Boashash, Evaluation of the modified S-transform for time frequency synchrony analysis and source localization, EURASIP J. Adv.
Signa. Process., 2012.
R.G. Stockwell, Why use the S-Transform?, AMS pseudo–differential operators: partial differential equations and time–frequency analysis 52
(2007) 279–309