See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263555142

Evaluating sewing thread consumption of jean pants using fuzzy and

regression methods

Article  in  Journal of the Textile Institute · October 2013

DOI: 10.1080/00405000.2013.773627

CITATIONS READS

13 3,273

2 authors:

Jaouachi Boubaker Faouzi Khedher

National School of Engineers of Monastir, Tunisia Engeneering loboratory textile University of Monastir TUNISIA
49 PUBLICATIONS   237 CITATIONS    25 PUBLICATIONS   97 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Qualité View project

wet-pneumatic spliced yarns View project

All content following this page was uploaded by Faouzi Khedher on 10 February 2016.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [Faouzi KHEDHER]
On: 04 April 2013, At: 06:58
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of The Textile Institute

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tjti20

Evaluating sewing thread consumption of jean pants

using fuzzy and regression methods
a a
Boubaker Jaouachi & Faouzi Khedher
a
Textile Engineering Laboratory of Ksar Hellal, University of Monastir, Monastir, Tunisia
Version of record first published: 15 Mar 2013.

To cite this article: Boubaker Jaouachi & Faouzi Khedher (2013): Evaluating sewing thread consumption of jean pants using
fuzzy and regression methods, Journal of The Textile Institute, DOI:10.1080/00405000.2013.773627

To link to this article: http://dx.doi.org/10.1080/00405000.2013.773627

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
 The Journal of The Textile Institute, 2013
http://dx.doi.org/10.1080/00405000.2013.773627

Evaluating sewing thread consumption of jean pants using fuzzy and regression methods
Boubaker Jaouachi* and Faouzi Khedher
Textile Engineering Laboratory of Ksar Hellal, University of Monastir, Monastir, Tunisia
(Received 8 December 2012; final version received 1 February 2013)

This study focuses on the evaluation of sewing thread consumption of jean pants using the fuzzy logic theory.
Referring to literature works, fuzzy logic method remains an accurate method that allows a new level of flexibility
over traditional mathematical methods in defining and evaluating constraints. The application of fuzzy rules and
fuzzy memberships is discussed and investigated. Using the influential parameters and optimized sewing condi-
tions (suitable adjusted regulations of each input) such as thread composition, needle size and fabric weight, the
results show that sigmoid membership gives better fitting of experimental results. Compared with the experimental
consumptions, theoretical findings of the jean pant can be predicted in the desired field of interest. The results also
indicate that the pant consumed thread remains influenced especially by the thread properties and the needle size
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

as well. Compared with regression model, the fuzzy model gives a more accurate prediction and provides widely
the amount of sewing thread than the regression model.
Keywords: consumption; sewing thread; jean pant; fuzzy; regression

Introduction standard for sewn product categories such as jean

Competitive markets require the reduction in expenses pants, shirts, and jacket. As a result, 10 to 15% of
to maximize profits. With recent increases in sewing wasted sewing thread is saved and added to the con-
speeds and with the advent of both synthetic fabrics sumption. This waste is caused by shop-floor condi-
and threads, the problem of excessive thread con- tions such as machine running, thread breakage,
sumption in sewing operations has become more criti- repairs, etc. Until now, there is no analytical model
cal and has occur a high interest. A practical and using fuzzy theory method to evaluate the amount of
reasonable estimation of potential sewing thread for sewing thread required to predict the sewing thread
garment industry will not only save money, but also consumption value. Compared with traditional mathe-
provide a better-quality of sewing thread for the same matical methods (regression, neural network, mathe-
cost. However, in the general literature survey, this matical, subjective, etc.) in defining, evaluating
problem has not been studied sufficiently for two rea- constraints, prediction and in modelling both complex
sons: (1) the complexity of the evaluation of thread and non-linear problems, fuzzy method offers
consumption and (2) the high input parameter large levels of flexibility (Altinoz & Winchester,
numbers. Regarding the literature, there are a limited 2001; Chang-Chiun & Wen-Hong, 2001; Hyung,
number of studies relating to thread consumption dur- Sung, Sook, Jae, & Seong, 2001; Jaouachi, 2010a,
ing sewed cloths. Although the thread consumption 2010b). In addition, the fuzzy logic method explores
evaluation using such techniques was measured as the possibility rather than the probability and as a
function of some input parameters such as the stitch consequence eliminates precision distorts. The purpose
length, thread tension and its compressive modulus of this study is to determine objectively the amount
(Amirbayat & Alagha, 1993; Dorrity & Olson, 1996; of sewing thread used by assembly type and finally
Jaouadi, Msahli, Babay, & Zitouni, 2006; O'Dwyer deduce the amount need by type of garment. There-
and Munden, 1975; Ukponmwan, Mukhopadhyay, & fore, it is focused on evaluation and prediction of
Chatterjee, 2000), the sewing thread consumption of thread consumption of jean pant using a database for
garment needs to be provided objectively and pre- fuzzy rules. The effectiveness of fuzzy model was,
dicted accurately. Nevertheless, these factors and oth- therefore, tested by comparing the predicted properties
ers are not constant with the different style of with both the experimental ones and with the
garment. That is why thread consumption is never regression analysis model results.

*Corresponding author. Email: boubaker_jaouachi@yahoo.fr

Copyright Ó 2013 The Textile Institute

 2 B. Jaouachi and F. Khedher

Fuzzy logic memberships and rules Table 1 illustrates the overall rules which are
In fuzzy logic, a fuzzy set contains elements with only maintained constant during the data collection due to
partial membership ranging from 0 to 1 to define both the optimized input and the output parameters
uncertainty for classes that do not have clear defined remain same during testing and training in the fixed
boundaries (Jaouachi, Ben Hassen, Sahnoun, & Sakli, field of interest. A Taguchi experimental design was
2010; Jaouachi, Louati, & Hellali, 2010; Majumdar, used because of the fact that building this database
Majumdar, & Sarkar, 2005). Generally, fuzzy model is consists in minimal (for the economic goals) and logic
built using four essential parts: fuzzification, a fuzzy experimental tests.
rule base, a fuzzy inference engine and defuzzification
(Chang-Chiun & Wen-Hong, 2001; Hyung et al.,
Materials and methods
2001). The first step, called fuzzification consists on
conversion of the feature values into appropriate fuzzy Data collection and analysis
sets. The second step tackled with the support of the Three different input parameters were chosen and used
fuzzy rule base. Hence, the fuzzified values are then in the experiment to sew jean pant samples and evalu-
inferred to provide decisions using the inference ate their experimental thread consumption. Table 1
engine. Consequently, the results are later converted shows the input parameters and their correspondent
into a crisp value by defuzzification. In the last step, levels. To regulate and adjust the splicing conditions,
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

the defuzzified value represents the decision made by three input parameters with two regulation points were
the fuzzy building model. In this work, four different varied. Hence, two different compositions of sewing
membership functions (Gaussian-shaped, trapezoidal, twisted thread (represented by Tc parameter) are 100%
sigmoid-shaped and triangular) are chosen and applied PES and 100% cotton. The twisted polyester and
as shown in Figure 1. cotton threads were prepared by twisting two and

Figure 1. Gaussian, trapezoidal, sigmoid-shaped and triangular membership used in our theoretical model.

Table 1. Fuzzy rules of sewing thread consumption, STC.

Rule no. Rules

1 If (Tc is Low) and (Ns is Low) and (Fw is (Low), then (STC is High) (1)
2 If (Tc is Low) and (Ns is Low) and (Fw is (High), then (STC is High) (1)
3 If (Tc is Low) and (Ns is High) and (Fw is (Low), then (STC is Medium) (1)
4 If (Tc is Low) and (Ns is High) and (Fw is (High), then (STC is High) (1)
5 If (Tc is High) and (Ns is Low) and (Fw is (Low), then (STC is Low) (1)
6 If (Tc is High) and (Ns is Low) and (Fw is (High), then (STC is Medium) (1)
7 If (Tc is High) and (Ns is High) and (Fw is (Low), then (STC is Medium) (1)
8 If (Tc is High) and (Ns is High) and (Fw is (High), then (STC is Medium) (1)
(1): Represents the weights that can be applied to each rule. In general, the specific weights should be from 0 to 1 under the weight setting.
 The Journal of The Textile Institute 3

three component yarns, respectively. The needle size studied. All adjustment conditions are regulated to
(represented by Ns parameter) as well as the fabric obtain good quality of assembly. Denim weaved fab-
weight parameter (Fw) is tested. There are two differ- rics were seamed on a JUKI DDL-8700 and MO-3316
ent compositions of the studied pant fabrics: 100% sewing machines with sewing needle Nm 90 and 120.
cotton and 98% cotton + 2% elastane. These optimized All machine settings were the same for every sample.
input parameters affect the sewing thread consumption Fabrics layers were seamed with stitch densities of 4.5
(STC) of the jean pant according to Taguchi analysis. and 3.5 stitches/cm which were not identical to those
The Taguchi experimental design is chosen to mini- selected by Webster et al. (1998).
mize objectively and keep the suitable numbers of Thus, this study is essentially carried out accord-
specimens. Each one of the two levels of the input ingly to these stitches. Overall seam operations are
parameters (I and II) for adjustment represents a regu- realized within the same type of mean sewing thread
lation point, with I corresponding to the minimum, count, 16.6 tex, to obtain a pant jean five pockets
and II referring to the maximum as shown in Table 2. (Figure 2). Thread consumption are measured by
The predicted output using fuzzy and regression unstitching the seam and measured accordingly to the
models are the sewing thread consumption (STC). French Standard NF G07 101. So, seam length was
As shown in Table 2, two types of jean fabrics measured after every operation of the garment
(heavy and light) have been chosen in the study. The (Table 2) and get thread requirement consumption of
majority of assemblies made at the jean pants sewing the garment by adding thread consumption of each
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

were sewn with different types of stitches: flat felled operation.

seam (2
401), safety stitch machine (ISO-516), over-
lockstitch machine (404) and lockstitch machine
(301). Their effects on the thread consumption will be Virgin fuzzy parameters
Sigmoid-shaped, triangular, Gaussian and trapezoidal
Table 2. Input parameters and their correspondent levels.
membership functions were applied. Referring to ear-
lier works (Jaouachi, Ben Hassen, et al., 2010; Jaou-
Twisted yarn Needle count Weight achi, Louati, & Hellali, 2010) and Majumdar’s results
Input level component, Tc (Nm), Ns (g/m2), Fw (Majumdar et al., 2005), the triangular shaped with
I 2 90 268 two membership membership functions for each input
II 3 120 417 gives an accurate prediction and a widely evaluation.
However, this finding remains relative of studied

401 - 401

401

504 + 301

516

301

Figure 2. Different stitch types in the pant or trouser sample.

 4 B. Jaouachi and F. Khedher

problem and functions of investigated situations and together when the twisted yarn number decreases (Fig-
can not be extrapolated in each case. For fuzzy model- ure 4). In addition, Figure 4 shows that the thread
ling, the Fuzzy Logic Toolbox of MATLAB software, composition becomes negligible when fabric mass
version 7.0.1 is used. Table 1 shows the fuzzy rules of value is near to 340 g/m2 and the needle size (Ns)
sewing thread consumption which are obtained by value is about Nm 105 as shown in Figure 5.
using previous experimental data and are trained for the To evaluate the effectiveness of sewing thread
fuzzy model. According to earlier works (Jaouachi, consumption after using fuzzy theory method (STCfuz),
Ben Hassen, et al., 2010; Jaouachi, Louati, & Hellali, the obtained values are compared to the actual or
2010), the basic structure of fuzzy logic model experimental ones (STCact). Table 3 shows the
includes fuzzying, fuzzy inference, fuzzy rule base and experimental database of the STC of pants and the
defuzzying was adopted. Fuzzy modelling method is errors (E) between the actual value and the value with
based on four different steps as defined below. fuzzy theory which are defined as:

(1) Defining input and output parameters: In this E(m) ¼ STCfuz  STCact : (1)
work, the sewing thread composition (Tc), nee-
dle size (Ns) and the fabric weight (Fw) are Referring to earlier work (Jaouachi, 2010b), these
analysed. Moreover, the sewing thread con- error values are the differences existing between
sumptions (STC) of pant are tested. experimental measured output values (STCact) and the
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

(2) Defining fuzzying conditions of parameters: theoretical ones (STCfuz) in each applied fuzzy mem-
Knowing the actual situation, it is possible to bership function case.
determine the universe of discourse of every Table 3 shows the overall tested pant specimen’s
input parameter. Overall input and output consumption and the variation between experimental
parameters as well as the scope of the operation (actual values) and theoretical values (predicted
are considered. The linguistic expressions and values) of the consumed thread which are expressed
their corresponding membership functions are
assigned.
(3) Designing fuzzy rules and fuzzy inference:
Linguistic fuzzy rules as presented in Table 1
are established according to the constructed
experimental database. Mamdani’s min–max
fuzzy inference approach (Cox, 1995) is
adopted to satisfy the requirements of the work.
(4) Selecting the fuzzying technique: The fuzzy out-
put conferred is converted into clear values and
centroid calculation method (returns the centre
of area under the curve of each output) is used
to defuzzify (Cox, 1995). There are five built-
in methods supported: centroid, bisector, med- Figure 3. Evolution of sewing thread consumption (STC) as
ium-maximum (the average of the maximum function of fabric weight, Fw and needle size (expressed by
value of the output set), high-maximum and Nm), Ns.
low-maximum.

Results and discussion

Figures 3–5 show the relationship among the effect of
the input parameters on the (STC) of pant samples.
From Figure 3 it is noted that, as the fabric weight
increases more than 330 g/m2, there is concomitant
increase in the consumption of thread.
The size of needle also exhibits similar influence
on sewing thread consumption of thread but from Nm
equals to 110. However, there is a strong relationship Figure 4. Evolution of sewing thread consumption (STC)
between the evolutions of consumption of thread for as function of thread composition, Tc (expressed by twisted
sewed pants and thread type because they increase thread number) and fabric weight, Fw.
 The Journal of The Textile Institute 5

developed fuzzy model, and by analogy with earlier

works evaluation results, it was concluded that the pre-
dicted values with fuzzy logic are better (R from the
regression model is 0.873 but values of R from the
fuzzy model are 0.95 for STC using sigmoid-shaped
function, 0.928 for STC using triangular-shaped func-
tion, 0.906 for STC using Gaussian-shaped function
and 0.904 for STC using trapezoidal-shaped function,
respectively). By comparison of the regression models
each others, given by fuzzy theory method, it is clearly
remarkable that the sigmoid-shaped and triangular-
Figure 5. Evolution of sewing thread consumption (STC)
as function of thread composition, Tc (expressed by twisted shaped (may be because it is formed with straight
thread number) and needle size (Nm), Ns. lines) membership functions give the best results.
Compared with the Taguchi analysis findings, the
by the error value. The STC is investigated and theoretical ones using fuzzy method prove, regarding
evaluated using fuzzy theory system. By comparing the Figures previously tackled, that all inputs are grad-
the theoretical output values with the experimental ually influents parameters on thread consumption of
pants. However, it is really notable that the number of
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

ones, the results show that the fuzzy model is verified

with tested data in the experimental design of interest. twisted yarns as well as the composition of sewing
After regression analysis results, by applied Taguchi thread (100% cotton or 100% PES) affects
design, the regression model shows non sufficient coef- enormously the predicted consumption values. Using
ficient of determination which equals to 0.763. More- PES thread count within three-twisted yarns increases
over, among the tested input parameters, only the the sewed pant consumption. These previous figures
thread composition one is important. This means that show that the needle size, Nm 105 value, is widely
the sewing thread composition seems the most influen- recommended using fuzzy theory to obtain the lowest
tial factor that affects the consumption of thread on sewing thread consumption of pant five pockets.
sewed pant specimens. The regression relationship
established, using all input parameters, is given by:
Conclusion
STC ¼ 552  148Tc  45; 2 Ns þ 28; 0 Fw : (2) STC of pant samples expressed with three input param-
eters and have been evaluated and predicted by fuzzy
Furthermore, to improve these findings, Figures system on this experimental field of interest. Moreover,
6–9 show the regression model evolutions of the STC this study reports the prediction of sewing thread
of the jean pant using fuzzy logic method. After consumption using a built fuzzy model. Compared
comparing the results given by the regression analysis with other experimental databases of input parameters,
model using statistical analysis (obtained by Taguchi the theoretical model can be used to predict the
analysis results) and those obtained using the consumed thread to sew pants. In fact, it was

Table 3. STC of pants predicted and evaluated by fuzzy theory.

Significant STCact. values (m)

inputs within Theoretical STCfuz values using Errors between fuzzified and
their levels fuzzy theory (m) actual values, E(m)
Tc Ns Fw Trimf Trapmf Sigmf Gaussmf ETrimf ETrapmf ESigmf EGaussmf
1 1 1 451.60 451 423 413 402 16.40 28.60 38.60 49.60
1 1 2 432.74 410 423 413 402 22.74 9.74 19.74 30.74
1 2 1 281.88 321 321 372 321 39.12 39.12 90.12 39.12
1 2 2 347.85 418 423 372 402 70.15 75.15 24.15 54.15
2 1 1 192.52 224 219 321 242 31.48 26.48 128.48 49.48
2 1 2 231.75 321 321 321 321 89.25 89.25 89.25 89.25
2 2 1 236.25 321 321 321 321 84.75 84.75 84.75 84.75
2 2 2 261.72 321 321 321 321 59.28 59.28 59.28 59.28
Note: STCact. represents the experimental STC values; ETrimf: error existing between the theoretical ones using Triangular membership function
and the experimental STC values; ETrapmf: error existing between the theoretical values using Trapezoidal membership function and the experi-
mental STC values; ESigmf error existing between the theoretical values using sigmoid membership function and the experimental STC values;
EGaussmf: error existing between the theoretical values using Triangular membership function and the experimental STC values.
 6 B. Jaouachi and F. Khedher

position which expressed by the twisted yarn number

and the fibre type (Tc), second by needle size (Ns) and
finally by fabric weight (Fw). The fuzzy forecast model
is developed according to the correlations between
inputs and output parameters as defined previously
using Mandani’s min-max inference. Experimental
findings are in good agreement with theoretical ones
regarding the classification of the most influential input
Figure 6. STC theoretical regression model evolution using parameter on the thread consumption. However, the
Sigmf membership function. developed fuzzy rules provide a better understanding
of the effects that input parameters have on sewing
thread consumption of jean pants. Further detailed
studies using other theoretical techniques will follow.

References
Altinoz, C. & Winchester, S. C. (2001). A fuzzy approach
to supplier selection. Journal of The Textile Institute, 92,
Downloaded by [Faouzi KHEDHER] at 06:58 04 April 2013

155–167.
Amirbayat, J. & Alagha, M. J. (1993). Further studies on
balance and thread consumption of lockstitch seams.
International Journal of Clothing Science and Technol-
Figure 7. STC theoretical regression model evolution using ogy, 5, 26–31.
Gaussmf membership function. Chang-Chiun, H. & Wen-Hong, Y. (2001). Fuzzy neural
network approach to classifying dyeing defects. Textile
Research Journal, 71, 100–104.
Cox, E. D. (1995). Fuzzy logic for business and industry
(pp. 39–59). Boston, MA: Charles River Media.
Dorrity, J. L., & Olson, L. H. (1996). Thread motion ratio
used to monitor sewing machines. International Journal
of Clothing Science and Technology, 8, 1–6.
Hyung, T. C., Sung, H. J., Sook, R. K., Jae, Y. J., &
Seong, H. K. (2001). Detecting fabric defects with
computer vision and fuzzy rule generation. Part II:
Defect identification by a fuzzy expert system. Textile
Research Journal, 71, 563–573.
Jaouachi, B., Ben Hassen, M., Sahnoun, M., & Sakli, F.
(2010). Evaluation of wet pneumatically spliced elastic
Figure 8. STC theoretical regression model evolution using denim yarns with fuzzy theory. Journal of The Textile
Institute, 101, 111–119.
Trimf membership function.
Jaouachi, B., Louati, H., & Hellali, H. (2010). Predicting
residual bagging bend height of knitted fabric using
fuzzy modelling and neural networks. AUTEX Research
Journal, 10, 110–115.
Jaouadi, M., Msahli, S., Babay, A., & Zitouni, B. (2006).
Analysis of the modeling methodologies for predicting
the sewing thread consumption. International Journal of
Clothing Science and Technology, 18, 7–18.
Majumdar, A., Majumdar, P. K., & Sarkar, B. (2005).
Application of an adaptive neuro-fuzzy system for the
prediction of cotton yarn strength from fibre properties.
Journal of The Textile Institute, 96, 55–60.
O’Dwyer, U. & Munden, D. L. (1975). A study of the factors
effecting the dimensions and thread consumption in 301
Figure 9. STC theoretical regression model evolution using seams – part I. Clothing Research Journal, 3, 3–32.
Trapmf membership function. Ukponmwan, J. O., Mukhopadhyay, A., & Chatterjee, K. N.
(2000). Sewing threads. Textile Progress Institute Jour-
concluded that the error values between analytical and nal, 30, 79–80.
Webster, J., Laing, R. M., & Niven, B. E. (1998). Effects of
experimental consumptions show the effectiveness of repeated extension and recovery on selected physical
the fuzzy model in prediction. With the help of the properties of ISO-301 stitched seams. Part I: Load at max-
fuzzy theory, the findings indicate that the thread imum extension and at break. Textile Research Journal,
consumption values are first influenced by thread com- 68, 854–864.

View publication stats