You are on page 1of 9

Original article

Textile Research Journal

82(15) 1560–1568

A study of processing parameters in ! The Author(s) 2012

Reprints and permissions:
open-end spinning process for organic DOI: 10.1177/0040517512441989
cotton (Ne 24)

Chung-Feng Jeffrey Kuo1, Wei Lun Lan2 and Feriyonika1

Organic cotton has become a popular product in recent years due to its environmental friendliness. Different from
conventional cotton, the planting of organic cotton does not use pesticides, herbicides, or fertilizer, which can cause
degradation in soil fertility, water and natural equilibrium. Organic cotton crop rotation can build strong soil, retain water
more efficiently, maintain a balance between pests and their natural predators, and control weed growth. This study
presented a yarn spinning process using organic cotton as the raw material. An open-end rotor spinning machine was
used in the spinning process. To analyze the processing parameters affecting yarn qualities, the Taguchi method and
response surface methodology were used to design an orthogonal array experiment and to find the relationship between
controllable input parameters (i.e. feed speed, winding speed, and rotor speed) and output qualities (i.e. strength,
unevenness, imperfection indicator/km, and hairiness). To find the optimal parameter combination, two popular opti-
mization methods, particle swarm optimization and genetic algorithm, were used. The results showed that with optimal
parameters (0.345 m/min, 34.717 m/min, 82,599 rpm for feed speed, winding speed, and rotor speed, respectively), the
proposed method can find optimum values in all observed qualities. A confirmation experiment was conducted to
validate the results.

Organic cotton, Taguchi method, response surface methodology, genetic algorithm, particle swarm optimization

parameters (i.e. feed speed, winding speed, and rotor

speed). The results were analyzed by analysis of variance
The conventional cotton planting process causes a ser- to determine significant parameters related to each qual-
ious environmental impact. Previous studies have sug- ity. Second, response surface methodology (RSM) was
gested that nearly all planting processes of conventional applied to use significant parameters to build relation-
cotton would damage the environment due to long term ships between parameters and each quality. Genetic
use of pesticides, herbicides, and fertilizer. The pollu- algorithm (GA) and particle swarm optimization
tion contaminates the air, soil and water, and further (PSO) were used to optimize all qualities. Finally, the
causes an unbalance in the biological and natural optimization results from both methods were evaluated
system. Different from conventional cotton, all planting based on strength quality (priority quality).
processes of organic cotton, from the seed preparation
to harvesting, are safe and environmentally friendly. It
can build strong soil through crop rotation, retain 1
Graduate Institute of Automation and Control, National Taiwan
water more efficiently, maintain a balance between University of Science and Technology, Taipei 106, Taiwan
pests and their natural predators, and control the 2
Department of Materials Science & Engineering, National Taiwan
weed growth through cultivation and hand hoeing.1 University of Science and Technology, Taipei 106, Taiwan
This study considered the qualities of the organic
cotton spinning process (i.e. strength, unevenness, Corresponding author:
Chung-Feng Jeffrey Kuo, Graduate Institute of Automation & Control,
imperfection indicator (IPI), hairiness), and used the L9 National Taiwan University of Science and Technology, 43, Keelung Road,
orthogonal array to reduce the number of experiments, Sec. 4, Taipei, 106, Taiwan, Republic of China
in order to combine the investigated processing Email:

Downloaded from by guest on July 9, 2016

Kuo et al. 1561

Open-end spinning process

During the 1960s, open-end spinning was one of the
new methods proposed to overcome speed and power
consumption problems in the ring spinning method for
yarn spinning technology. This method separates yarn
twisting action and winding action, and rotates the
package winding action at the relatively low winding
In the open-end spinning, various methods have
been invented, which can be classified into five
groups: vortex assembly, axial assembly, discontinuous
assembly, friction spinning and rotor spinning. This
study used the rotor spinning method, as shown in
Figure 1.2 Some studies have analyzed and optimized
the behavior of rotor spinning method related to the
quality of resulted yarn. Yang3 et al. used a linear
dynamic model to stabilize the motion of the rotor-
spun composite yarn spinning process, and used Figure 1. The rotor spinning.
approximate oscillating frequencies in vertical and hori-
zontal directions to find the optimal construction mix-
ture. Cheng4 et al. investigated the effects on properties
of open-end cover spun yarns by changing the setting of Table 1. Investigated parameters
yarn construction and formation. Pouresfandiari5 et al.
investigated a new method of combining filament yarns Design of experiment
and staple fibers on a modified open-end rotor spinning Feed speed Winding speed Rotor speed
frame. They produced three kinds of hybrid yarns by Level (m/min) (m/min) (rpm)
changing filament over-feed. Kuo6 et al. investigated
processing parameters affecting multiple qualities of 1 0.315 34.705 80000
bamboo charcoal and CVC (cotton 70%/ polyester 2 0.330 35.294 84000
30%) blended fibers. In their work, a combination of 3 0.345 35.882 88000
feed speed, winding speed, and rotor speed were
observed, and multiple qualities of bamboo charcoal
and CVC were also observed. In this study, the com-
bination of these three factors was investigated to find to combine the investigated parameters (three factors –
optimal qualities for organic cotton (Ne 24). three levels), as shown in Table 1.

Quality measurement and experimental

Analysis methods
The qualities (i.e. strength, unevenness, IPI/km, and
Analysis of variance (ANOVA)
hairiness) were evaluated to determine the combination After all the experiments were completed, the ANOVA
of processing parameters (feed speed, winding speed, was applied to analyze significant parameters related to
and rotor speed) with the most significant effect. In each quality. Details of the ANOVA are presented in
the experiment, the raw material was 100% organic the following subsection, as expressed in equation 1
cotton sliver, and an Uster Quickspin open-end rotor to 8.7
spinning experimental machine was used to spin open-
end spun yarn. The opening roller revolution was a. The degree of freedom (DF) was defined as:
8000 rpm, the rotor model was 31-N3; the yarn strength
was tested by a universal tension tester, Orientec DF ¼ r  1 ð1Þ
Tension RTA-1 T; and the evenness was tested by an
Uster Test-3 yarn evenness tester. Test temperature was where r is the number of experiments.
maintained at 20 C2 C, and relative humidity was
kept at 65%2%. The L9 orthogonal array was used b. Total sum of squares (TSS):

Downloaded from by guest on July 9, 2016

1562 Textile Research Journal 82(15)

TSS ¼ y2j  CF ð2Þ
P 2
j¼1 yj
CF ¼ ð3Þ

yj is the S/N ratio of experimental observation; n is

the number of experiments, and CF is the correc-
tion factor.

c. Sum of squares (SS): Figure 2. General procedure in genetic algorithm.

P hPn n
i¼1 j¼1 yij
SS ¼  CF ð4Þ

d. Error sum of squares (SSe)

SSe ¼ TSS  SSP ð5Þ

SSe is the error sum of square and P is the number

of the factor.

e. Variance (V) stands for the sum of the square

divided by the degree of freedom:
SS Figure 3. Direction of the velocity and position updates in
V¼ ð6Þ PSO.

f. Error mean sum of square (Ve):

SSe Table 2. Parameters of genetic algorithm used in this work
Ve ¼ ð7Þ
Population 60

With DFE ¼ (Total number of trials  number of Percent of crossover 85%

repetitions  1)  (DF). Type of crossover Two point crossover
Percentage of mutation 1%
g. Percent contribution () Generation 100–1000
Length of each chromosome 23 bit
SS  DF  Ve
¼  100% ð8Þ Number of chromosome 3 (1, 2, 3)

Genetic Algorithm (GA) Table 3. Parameters of particle swarm optimization

The particle swarm optimization is a search heuristic Population 60

that mimics the process of natural evolution. This heur- Inertia factor (w) Random (0.4–1.4)
istic is routinely used to generate useful solutions to Self confidence (c1) Random (1.5–2)
optimize and search problems.8 The main components Swarm confidence (c2) Random (2–2.5)
of a GA are encoding, fitness function, crossover,
Generation 100–1000

Downloaded from by guest on July 9, 2016

Kuo et al. 1563


Collecting result of experiment using L9 orthogonal array for all qualities

Build second order RSM equation for each quality based on results of ANOVA

Integrated all RSM equations as fitness function of PSO and GA

Applying PSO and GA to find optimal parameters

Finding optimum



Figure 4. Flow diagram of the proposed method.

Table 4. Experiment results with four repetitions

Particle swarm optimization (PSO)
Particle swarm optimization is inspired by the ability of
Strength Unevenness Imperfection Hairiness
flocks of birds, schools of fish and herds of animals to
(kgf) (%) (per km) (hairs/m)
adapt their environment. There are three main steps in
0.27925 12.535 75.75 5.3025 PSO: generating positions and velocities of particles,
0.27925 13.0525 104.25 5.1775 velocity update, and position update.9,10 Detail of
0.212 13.01 90.5 5.385 steps are briefly described in following subsection.
0.30175 12.9225 104.75 5.32
0.311 13.2425 144 5.3325 1. First, defining initialization positions, xik , and velo-
0.29475 13.045 78.75 5.49
cities, vik as expressed in equation 9 and 10.
0.30175 13.055 85.75 5.6725
xi0 ¼ xmin þ rand ðxmax  xmin Þ ð9Þ
0.30725 12.8525 60.75 5.7
0.26725 12.7625 68.5 5.645 xmin þ rand ðxmax  xmin Þ position
vi0 ¼ ¼ ð10Þ
t time

2. The second step is to update the velocities of all par-

ticles at time k + 1. The three values that affect the
mutation, population, and selection models. The gen- new search direction, namely, current motion, par-
eral procedure is depicted in Figure 2. ticle own memory, and swarm influence, are incor-
In a GA, a potential solution of a problem is pre- porated via a summation approach as shown in
sented for each chromosome where an evolution pro- equation 11 with three weight factors: inertia
cess run on a population of chromosomes corresponds factor, w; self confidence factor, c1; and swarm con-
to a search through a potential solution. This algorithm fidence factor, c2.
is usually used in global minimum or optimum solu- p  xik
i i
tions to a function which has many local minimums vkþ1 ¼ wvk þc1 rand
|{z} t
or optimums such as the Schaffer function, Rastrigin current
motion particle memory influence
function, Rosenbrock function and Griewank function.  
For those functions, GA has proved that it can find the pgk  xik
þ c2 rand ð11Þ
best solution from many traps of local minimums or t
maximums. swarm influence

Downloaded from by guest on July 9, 2016

1564 Textile Research Journal 82(15)

Table 5. ANOVA of strength and unevenness (strength/unevenness)

Source DOF SS Variance F-Ratio SS’ Percent contribution

A 2/2 3.4390/ 0.0346 1.7195/ 0.0173 88.06328/ 16.6949 3.400/ 0.0325 40.4423/ 21.0879
B 2/2 3.3895/ 0.0306 1.6947/ 0.0153 86.7955/ 14.7733 3.3504/ 0.0285 39.8534/ 18.5061
C 2/2 1.0122/ 0.0589 0.5061/ 0.0295 25.9191/ 28.4577 0.9731/ 0.0568 11.5753/ 36.8927
Error 29/29 0.5662/ 0.0300 0.0195/ 0.0010 – – 8.1290/ 23.5133
Total 35/35 8.4069/ 0.1540 – – 7.5911/ 0.1178 100/ 100

Table 6. ANOVA of IPI and Hairiness (IPI/hairiness)

Source DOF SS Variance F-ratio SS’ Percent contribution

A 2/2 18.4843/ 0.6407 9.2421/ 0.3204 223.2005/ 1426.8696 18.4015/ 0.6403 46.0813/ 85.3715
B 2/2 4.8815/ 0.0480 2.4407/ 0.0240 58.9445/ 106.8961 4.7987/ 0.0476 12.0169/ 6.3403
C 2/2 15.3661/ 0.0548 7.6830/ 0.0274 185.5479/ 121.9278 15.2833/ 0.0543 38.2726/ 7.2403
Error 29/29 1.2008/ 0.0065 – – – 3.6293/ 1.0478
Total 35/35 39.9326/ 0.7500 – – 38.4834 100/ 100

The tuning of the PSO algorithm weight factors is a where " is a random error. The  coefficients, which
topic that warrants proper investigation but is out- should be determined in the second-order model, are
side the scope of this work. For optimization par- obtained by the least square method.
ameters investigated in this work, the weight factors In this work, PSO and GA were programmed in
use the random values, 0.4 to 1.4, 1.5 to 2, 2 to 2.5 MATLAB with the following computer specification:
for w, c1 , and c2 respectively. Windows operating system and 2 GB of memory. The
3. The last step is the position update. The position of input parameters were encoded as X1, X2, and X3, for
each particle is updated using its velocity vector as feed speed, winding speed, and rotor speed, respect-
expressed in equation 12 and depicted in Figure 3. ively. The qualities were encoded as Y1, Y2, Y3, and
Y4 for strength, unevenness, imperfection indicator,
xikþ1 ¼ xik þ vikþ1 t ð12Þ
and hairiness, respectively. The solutions were bounded
in the following conditions: 0.315  X1  0.345,
34.705  X2  35.882, and 80000  X3  88000. Details
of properties are presented in Tables 2 and 3.
Proposed analysis method Figure 4 shows the flow diagram of proposed method.
To optimize all qualities of organic cotton spun yarn,
PSO and a GA were employed. These two optimization
methods aimed to find as high as possible values for Results and discussion
strength and as low as possible for unevenness, IPI, and Analysis of variance (ANOVA) and response surface
hairiness. The analysis firstly applied ANOVA to find
the most significant processing parameters related to
each quality. Second-order RSM, as expressed in equa- The investigation was accomplished with four repeti-
tion 13, was employed to build a relationship between tions. Table 4 shows the average of results based on
significant processing parameters and each quality. an L9 orthogonal array design. To find significant fac-
Four equations derived from RSM were subsequently tors related to each quality, ANOVA was employed,
integrated to obtain a fitness function which will be with the results shown in Table 5 and 6. As seen, all
optimized by both optimization methods. processing parameters showed significant effects on all
qualities. Based on this result, all parameters were
k X
k included to build the second-order response surface
y ¼ 0 þ i xi þ ii x2i methodology in equation 13.
i¼1 i¼1
XX For the RSM analysis, the results are expressed in
þ ij xi xj þ " ð13Þ equations 14 to 17. The R-squared (R2) values of the
i j models are 99.78%, 99.95%, 99.97%, and 99.84% for

Downloaded from by guest on July 9, 2016

Kuo et al. 1565

Table 7. Results of genetic algorithm

Fitness Elapsed time Y1 Y2 Y3 Y4

Iteration value (second) (strength) (um%) (IPI) (hairiness) X1 X2 X3

100 89.05 17.89 0.28846 12.869 68.658 5.6474 0.345 34.712 86324
200 63.38 35.87 0.27389 12.606 43.233 5.6788 0.345 34.710 83311
400 77.42 70.29 0.28132 12.744 57.156 5.6502 0.345 34.718 85013
600 58.91 108.37 0.27244 12.563 38.792 5.6959 0.345 34.717 82599
800 90.21 144.02 0.28918 12.885 69.796 5.6493 0.345 34.706 86484
1000 84.71 180.20 0.28562 12.824 64.363 5.6475 0.345 34.711 85879

Table 8. Results of particle swarm optimization

Fitness Elapsed time Y1 Y2 Y3 Y4

Iteration value (second) (strength) (um%) (IPI) (hairiness) X1 X2 X3

100 86.77 0.18 0.28045 12.93 66.331 5.6492 0.345 35.882 81964
200 43.10 0.50 0.26871 12.431 23.03 5.8045 0.345 34.705 80000
400 86.77 0.75 0.28045 12.93 66.331 5.6492 0.345 35.882 81964
600 86.77 1.12 0.28045 12.93 66.331 5.6492 0.345 35.882 81964
800 86.77 1.48 0.28045 12.93 66.331 5.6492 0.345 35.882 81964
1000 86.77 1.98 0.28045 12.93 66.331 5.6492 0.345 35.882 81964

Y1 (strength), Y2 (unevenness), Y3 (IPI), and Y4 (hairi-

ness), respectively.

Y1 ¼ 90:969 þ 4:6775X2 þ 2:2006  104 X3

 80:094X21  6:3986  102 X22 þ 5:3536
 1010 X23 þ 1:5833X1 X2  2:991
 105 X1 X3  8:5209  106 X2 X3 ð14Þ

Y2 ¼ 567:92 þ 22:289X2 þ 4:4784  103 X3

Figure 5. Fitness value in different generations.  198:69X21  0:21562X22 þ 5:5986
 109 X23 þ 10:297X1 X2  2:8729
 103 X1 X3  1:2532  104 X2 X3 ð15Þ

Y3 ¼ 5:5936  104 þ 2778:7X2 þ 0:17051X3

 9:8296  104 X21  43:313X22 þ 3:8311
 107 X23 þ 2126:9X1 X2  0:13356X1 X3
 5:2156  103 X2 X3 ð16Þ

Y4 ¼ 249:42  10:538X2  1:4621  103 X3

þ 351:02X21 þ 0:16201X22 þ 4:3119
 109 X23  6:9371X1 X2 þ 3:2549
Figure 6. Output of optimal parameters using GA.  104 X1 X3 þ 1:7520  105 X2 X3 ð17Þ

Downloaded from by guest on July 9, 2016

1566 Textile Research Journal 82(15)

Figure 7. Output of optimal parameters using PSO.

Figures 5 to 7 show the fitness values and outputs of

both methods. In this work, by analyzing the fitness
values, except in 200 iterations, PSO was trapped in
local maximum. Conversely, GA, with its operator
such as mutation, can escape from local maximum to
other solutions. This condition is a special advantage to
provide alternative solutions in engineering practice
level. In the next analysis, the most important quality
(strength) will be considered as a key factor to deter-
mine the solution.
To choose the best solution, the output of all quali-
ties in Figures 6 and 7 are analyzed. The chosen gener-
ation for each method is based on best fitness value:
Figure 8. Comparison of output values.
600th generation for GA and 200th generation for PSO.
The comparison among GA, PSO, and the set of best
Genetic algorithm (GA) and particle swarm single qualities is shown in Figure 8 and Table 9.
For the first quality, strength, the result of GA is
optimization (PSO)
stronger than PSO, 0.27244 to 0.26871. Both values
In the optimization process, the fitness function, equa- are still below the best single observed data, 0.3. If
tion 18, was derived by adding equation 14 to 17. compared to the average strength of observed data,
Because fitness values were evaluated based on the 0.2838, the optimal parameter is 4% less for GA and
maximum value, the Y2, Y3, and Y4 values, which are 5.31% less for PSO. This condition is still within toler-
minimum functions, were changed to minus signs. The ance because of other quality consideration. In terms of
results of the optimization are provided in Tables 7 the second quality, unevenness, the result of PSO is
and 8. better than GA and the observed data with values of
12.5, 12.4, 12.5 for GA, PSO and observed data,
respectively. Although the result of PSO is better than
max f ðx1 , x2 , x3 Þ ¼ Y1  ðY2 þ Y3 þ Y4 Þ
GA, the result of GA is still the same as minimum
¼ 5:6163  104  2:7858  103 X2 observed data.
 1:7331  101 X3 þ 9:8063 PSO leads over GA in terms of the third quality, IPI.
However, both are better than the minimum observed
 104 X21 þ 4:3303  10X22 data: 38.792, 23.03, 68.5 for GA, PSO and observed
 3:9249  107 X23  2:1287 data, respectively. For the last quality, hairiness, the
 103 X1 X2 þ 1:3607  101 X1 X3 result of GA is stronger than PSO, 5.6959 compared
to 5.8045. Both of them are still worse than the best
þ 5:3148  103 X2 X3 ð18Þ single observed data, 5.1775. If compared to the

Downloaded from by guest on July 9, 2016

Kuo et al. 1567

Table 9. Comparison of output values and best coordinate solution

Y1 (high Y2 (low Y3 (low Y4 (low

is better) is better) is better) is better) X1 X2 X3

GA 0.27244 12.5 38.792 5.6959 0.345 34.717 82599

PSO 0.26871 12.4 23.03 5.8045 0.345 34.705 80000
Best from 0.30725 12.5 68.5 5.1775 – – –

Table 10. Confirmation experimental results Table 11. Optimal value and standards

Strength Um IPI Hairiness Standards Optimal value

No. (kgf) (%) (per km) (hairs/m)
Strength (kgf) 24 Ne 0.252 kgf 0.297 kgf
1 0.311 12.62 41 5.53 Um (%) Less than 10.8% 12.71%
2 0.285 12.58 62 5.55 IPI (per km) Less than 107 per km 48.8 per km
3 0.314 12.54 58 5.59 Hairiness (hairs/m) No fixed standards 5.49 (hairs/m)
4 0.276 12.60 39 5.52
5 0.297 13.20 44 5.26
Average 0.297 12.71 48.8 5.49 where Z is the average value and Z^ is the S/N average.
S/N ratio 10.5886 22.2083 33.9256 14.7934

1 1
CIsn ¼ F;1,v2  Ve  þ ð20Þ
neff r
average hairiness of observed data, 5.45, the optimal
parameter is 4.56% and 6.56% higher, respectively where Fa;1,v2 is the F-value with significance level a, v2
for GA and PSO. is the degree of freedom of the pooled error variance,
Both optimization methods can find high optimal Ve is the pooled error variance, neff is the number of
solutions in term of all qualities. PSO can lead in valid observations, and r is the number of observations
term of Y2 and Y3, but not for others. Conversely, used to calculate the mean.
GA shows stable performance in all qualities. It can
reach a better quality of strength than PSO, but is Z^  CIsn 5 mZ 5 Z^ þ CIsn ð21Þ
not too weak for other qualities. Based on engineering
judgment (considering the most important quality, where Z^ is the S/N ratio average and mZ is the actual
strength),9 the result of GA is chosen as the solution mean value in the confirmation experiment.
of this problem. The optimum parameters are 0.345 m/ Hence, S/N ratio confidence intervals are,
min, 34.717 m/min, and 82,599 rpm for feed speed, 12.9394 < mstrength < 6.1179, 22.9913 < mum%
winding speed, and rotor speed, respectively. < 21.0688, 41.9179 < mIPI < 26.3015, and
14.9903 < mhairiness < 13.6761; S/N ratios of all
quality characteristics fall within the 95% CI, indicat-
Confirmation experimental ing that this experiment is reproducible and reliable.
This study conducted five verification tests on open-
Verification testing was conducted to verify whether the
end spun yarn multiple quality optimal parameter com-
mean under the estimated optimal condition is valid, in
bination, as shown in Table 10.
order to determine if data analysis conclusion is correct.
The confidence interval of one factor level mean was
calculated. This study conducted an estimation by Conclusion
using the S/N ratio of the combination of greatest
In this work, the parameter design of the Taguchi
impact factor and optimal process parameter. The
method, RSM and optimization methods, GA and
95% confidence interval of its S/N ratio is calculated
PSO, were employed to find the optimal processing par-
as follows:
ameters for organic cotton spun yarn. ANOVA was
conducted to determine the significant parameters for
Z^ ¼ Z þ ðA2 ZÞ þ ðB3 ZÞ þ ðC1 ZÞ ð19Þ each response. The RSM was then used to build the

Downloaded from by guest on July 9, 2016

1568 Textile Research Journal 82(15)

relationship between input parameters and output Funding

responses. The relationship between them is described This work was supported by the National Science Council of
by RSM’s quadratic model. the Republic of China (grant number 97-2221-E-011-030-
The optimal parameters are: feed speed of MY3).
0.345 m/min, winding speed of 34.717 m/min and
rotor speed of 82,599 rpm. The analysis of two opti-
mization methods indicated that both methods can
reach a high performance in terms of fitness value References
and output of each quality. The chosen solution was 1. Eyhorn F, Ramakrishnan M and Mader P. The viability
determined based on engineering judgment by consider- of cotton-based organic farming systems in India. Int J
ing the most important quality, yarn strength. In terms Agri Sustainability 2007; 5: 25–38.
of computational effectiveness, it was confirmed that 2. Oxtoby E. Spun yarn technology. London: Butterworth &
Co (Publishers) Ltd, 1987.
PSO has less consuming time than GA.
3. Yang RH and Wang SY. A linear dynamic model for
To validate the optimization results, confirmation
rotor-spun composite yarn spinning process. J Phys
experiment was conducted. Verification testing found Conf Series 2008; 96(1): 012039.
that, the S/N ratios of all the quality characteristics 4. Cheng KB and Murray R. Effects of spinning conditions
fall within the 95% confidence interval, proving that on structure and properties of open-end cover-spun
this experiment has reproducibility and reliability. yarns. Textile Res J 2000; 70(8): 690–695.
Moreover, as proved by the experimental results, 5. Pouresfandiari F, Fushimi S, Sakaguchi A, et al.
when applied to an open-end spinning system, the opti- Spinning conditions and characteristics of open-end
mal parameter combination can achieve good strength rotor spun hybrid yarns. Textile Res J 2002; 72(1): 61–70.
and even spinning quality, which meets the expected 6. Kuo CFJ, Wei HJ, Huang CC, et al. Processing param-
objectives of this study. Good single yarn strength eters optimization of multiple quality characteristics of
(0.297 kgf), um% (12.71%), IPI (48.8/per km), and open-end rotor spinning process for bamboo charcoal
hairiness (5.49 hairs/m) can be obtained, where single and CVC blended fibers. Fiber Polym 2010; 11: 891–898.
yarn strength and IPI are better than that of an equiva- 7. Roy R. A premier on the Taguchi Method. Michigan:
Society of Manufacturing Engineering, 1990.
lent combed yarn, as shown in Table 11. Through opti-
8. Genetic Algorithm.
mal parameter design, it can elevate organic cotton
Genetic_algorithm (accessed December 5th, 2011).
yarn’s quality effectively, even better than combed 9. Hassan R, Cohanim B and de Weck O. A comparison of
yarn. It shows that through optimal parameter particle swarm optimization and genetic algorithm.
design, organic cotton can substitute cotton yarn use American Institute of Aeronautics and Astronautics,
in low counts spinning. It can decrease the use of pesti- 2004.
cides, herbicides, and fertilizer, and help environmental 10. Cleck M. Particle swarm optimization. London: ISTE,
protection. 2006.

Downloaded from by guest on July 9, 2016