Application of artificial neural network (ANN) for prediction of fabrics ...

Fibers and Polymers 2010, Vol.11, No.6, 917-923

DOI 10.1007/s12221-010-0917-8

Application of Artificial Neural Network (ANN) for Prediction of Fabrics’ Extensibility Tomislav Rolich, Anica Hursa Šajatoviéc*, and Daniela Zavec Pavliniéc1

Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipoviéca 28a, HR-10000 Zagreb, Croatia 1 Biomed d.o.o., Tugomerjeva 2, SI-1 000 Ljubljana, Slovenia

(Received January 12, 2010; Revised May 4, 2010; Accepted May 7, 2010)

Abstract: In the field of clothing technology, prediction of the fabric properties is very important because the fabric is the

basic element of every clothing item. Knowing the fabric properties it is possible to predict fabrics' behaviour during process of clothing manufacturing (in phase of cutting, sewing and ironing) as well as clothing items' behaviour during usage. According to the fabrics’ characteristics and model design it is possible to predict appearances of the clothing items and their draping which can be presented with many computer simulations. In this paper extensibility of the fabric which appears during a small forces loading on the fabrics are investigated. Loading of small forces on the fabric appears in each phases of clothing manufacturing processes and during usage of clothing items. Investigations are managed on 50 fabrics which are weaving in twill weave and 100 % wool. The basic characteristics of fabric (density of warp and weft, mass per unit area, thickness) are defined according appropriate standard methods and tensile properties in the warp and weft directions are measured using KES-FB1 measuring system. Using an artificial neural network (ANN) prediction of extensibility properties of the fabrics are done, results are compared with experimental values and deviations are determined. ANN is an adaptive system that changes its structure based on external or internal information that flows through the network during the learning phase. They can be used to model complex relationships between inputs and outputs or to find patterns in data. Based on the implemented investigations, minimal deviations between experimental and predicted values are obtained and can be concluded that ANN can be used for prediction of the fabrics properties. Keywords: Clothing technology, Fabric, Extensibility, Artificial neural network (ANN)

properties on clothing item appearance [12-15]. According to mentioned, knowing of fabric properties are important so in the paper is investigated application of the artificial neural network (ANN) for predicting of fabric tensile properties.

Introduction

Knowing of the fabric properties is important in garment manufacturing processes and during usage of garments. In the phase of clothing design it is important to know all material properties for prediction of appearance and draping of garment. Today on the market it can be find many software which have possibilities very truthfully to show the appearance and draping of clothing based on the model design and material characteristics [1-4]. In garment manufacturing processes, on fabrics that are the fundamental part of each garment different types of forces have influence on manufacturing processes. Manufacturing processes and behaviour of garments during usage (wearing and washing) depend on fabric’s properties. Many scientists are investigated fabric properties [5-9], and based on comprehensive and intensive investigations they constructed two systems for objective measuring of the physical and mechanical fabric properties: FAST (Fabric Assurance by Simple Testing) and KES (Kawabata Evaluation System). Using mentioned measuring systems it can be possible to exactly determine many fabric properties which indicate on phenomena of some problems in garment manufacturing processes, respectively their appearance. Some scientists are investigated problems which appears in garment manufacturing processes and manner how to avoid and remove them [10,11], and some are investigated the influence of fabric

Materials and Methodology

Applications of the artificial neural network (ANN) for predicting of fabric tensile properties (extensibility) are investigated in the paper. Values of experimental measurement of extensibility using KES-FB1 measuring device are compared with predicted values that are computed with ANN. The measuring system, exactly measuring device KES-FB1, the artificial neural network and fabric characteristics are described separately. In Figure 1 the plan of investigation is shown.

Figure 1. Plan of investigation.

*Corresponding author: [email protected] 917

918 Fibers and Polymers 2010, Vol.11, No.6

KES-FB1 measuring device for tensile and shearing testing; (a) KES-FB1 testing device and (b) original layout of the results. Figure 2.

KES-FB Measuring System

KES-FB measuring system is used for objective evaluation of the physical and mechanical fabric properties. Measuring system consists of four measuring device [16]: · KES-FB1 measuring device for tensile and shearing testing, · KES-FB2 measuring device for bending testing, · KES-FB3 measuring device for compression testing and · KES-FB4 measuring device for surface testing. KES-FB1 measuring device (Figure 2) is used for tensile and shearing testing of fabric properties. The test specimen 20 cm wide is fixed in two clamps and exposed to the uniaxial tensile loading. One clamp is fixed and the other is moving with constant velocity until the maximum value of force is reached Fm =490.35 cN/cm. After that the moving clamp starts moving in the opposite direction, which results in unloading of the specimen. Results of the measurement consist of the plot load vs. deformation, extensibility (EMT in %) and additional data on deformation work (WT in cNcm/cm2), linearity (LT) and relaxation capacity (RT in %) [16].

Artificial Neural Network

An artificial neural network (ANN), usually called neural

Tomislav Rolich et al.

network (NN), is a mathematical model or computational model that tries to simulate the structure and/or functional aspects of biological neural networks [17]. It consists of an interconnected group of artificial neurons and processes information using a connectionist approach to computation. In most cases an ANN is an adaptive system that changes its structure based on external or internal information that flows through the network during the learning phase. Neural networks are non-linear statistical data modelling tools. They can be used to model complex relationships between inputs and outputs or to find patterns in data. Although computing these days is truly advanced, there are certain tasks that a program made for a common microprocessor is unable to perform; even so a software implementation of a neural network can be made with their advantages and disadvantages. Advantages: · A neural network can perform tasks that a linear program cannot. · When an element of the neural network fails, it can continue without any problem by their parallel nature. · A neural network learns and does not need to be reprogrammed. Disadvantages: · The neural network needs training to operate. · The architecture of a neural network is different from the architecture of microprocessors therefore needs to be emulated. · Requires high processing time for large neural networks. Another aspect of the artificial neural networks is that there are different architectures, which consequently requires different types of algorithms, but despite to be an apparently complex system, a neural network is relatively simple. In engineering, neural networks serve two important functions: as pattern classifiers and as nonlinear adaptive filters. An artificial neural network is an adaptive, most often nonlinear system that learns to perform a function (an input/ output map) from data. Adaptive means that the system parameters are changed during operation, normally called the training phase. After the training phase the artificial neural network parameters are fixed and the system is deployed to solve the problem at hand (the testing phase). The artificial neural network is built with a systematic stepby-step procedure to optimize a performance criterion or to follow some implicit internal constraint, which is commonly referred to as the learning rule. The input/output training data are fundamental in neural network technology, because they convey the necessary information to “discover” the optimal operating point. The nonlinear nature of the neural network processing elements provides the system with lots of flexibility to achieve practically any desired input/output map. An input is presented to the neural network and a corresponding desired or target response set at the output.

Fibers and Polymers 2010, Vol.11, No.6

Application of ANN for Prediction of Fabrics' Extensibility

An error is composed from the difference between the desired response and the system output. This error information is fed back to the system and adjusts the system parameters in a systematic fashion (the learning rule). The process is repeated until the performance is acceptable. If one does not have data that cover a significant portion of the operating conditions or if they are noisy, then neural network technology is probably not the right solution. On the other hand, if there is plenty of data and the problem is poorly understood to derive an approximate model, then neural network technology is a good choice. ANN-based solutions are extremely efficient in terms of development time and resources, and in many difficult problems artificial neural networks provide performance that is difficult to match with other technologies. ANN method was implemented in the field of textile and garment technologies few years ago. Some authors applying discriminant analysis and neural network successfully to establish discriminant models for fabric characteristics of cotton, linen, wool and silk. The characteristics and properties of mentioned fabric was measured using KES-FB system and FAST system [18]. Beltran [19] were developed a method for predicting worsted spinning performance with an artificial neural network (ANN) trained with backpropagation. Bevilacqua [20] were used a neural network approach for defects identification in textile. The images analyzed came from an artificial vision system that they used to acquire and memorize them in bitmap file format. The artificial neural networks (ANN) are manipulated to compress a bitmap that may contain several defects in order to represent it with a number of coefficients that is smaller than the total number of pixel but still enough to identify all kinds of defects classified. Some researches are investigated application of ANN to predict of sewing performance of fabrics. The purpose of this study is to verify the ANN techniques that could be emulated as human decision in the prediction of sewing performance of fabrics [21]. The other researchers investigate the possibility of using counterpropagation neural networks to identify the combinations of dyes in textile printing paste formulations [22]. et al.

et al.

Characteristic of Materials

During the investigation of application of ANN for predicting the fabric extensibility, 50 different fabrics are used. The fabrics were weaved in twill and made of 100 % wool [15]. Characteristics of the fabrics as density of warp and weft (determined using HRN EN 1049-2:2003 standard), mass per unit area (determined using HRN ISO 3801:2003 standard), thickness (determined using KES-FB3 measuring device) and extensibility in warp (EMT1) and weft (EMT2) direction (determined KES-FB1 measuring device) are shown in Table 1. In this paper, fabrics for women jackets and costumes were investigated [15]. Selection of investigated fabrics is

919

Characteristics of the fabrics (input for ANN) and extensibility (output for ANN) measured using KES-FB1 measuring device Density Mass per Thickness EMT1 EMT2 Mark of Density of warp of weft unit area (mm) (%) (%) fabric (yarn/cm) (yarn/cm) (g/m2) Table

1.

TK 555 TK 150 TK 166 TK 524 TK 556 TK 523

TK 482

TK 076 TK 090 TK 466 TK 522 TK 469 TK 520 TK 165 TK 091 TK 581

TK 258

34 25 30 25 35 26 16

16 33 30 24 30 22 30 33 15 28

23 25 25 25 25 22 16

14 20 25 22 25 22 28 20 15 21

TK 111 TK 256 TK 257 TK 317 TK 093 TK 316 TK 096 TK 049 TK 467 TK 100 TK 102 TK 047 TK 041 TK 110 TK 338 TK 101 TK 415 TK 307

44 28 28 9 46 8 9 27 50 9 9 27 11 9 9 9 10 9

28 21 21 8 29 7 9 24 30 9 9 24 10 9 8 9 10 8

TK 050

27

24

TK 098 TK 334 TK 323 TK 333 TK 455 TK 048 TK 259 TK 416 TK 417 TK 306 TK 260 TK 474 TK 473 TK 527

5 9 10 9 12 27 28 30 25 10 28 40 28 70

4 8 9 8 10 24 25 30 30 10 25 30 30 46

169.59 179.50 148.61 170.32 173.27 165.55 211.37

209.50 189.10 212.38 160.31 221.36 169.97 148.45 198.50 252.22 237.23

272.40 257.14 250.79 303.30 270.10 289.00 317.60 223.20 270.30 269.90 276.80 224.80 262.30 260.10 346.00 294.50 289.22 344.30 224.40

413.20 336.20 338.80 343.20 280.52 223.30 194.45 211.36 208.71 348.50 210.44 210.10 213.06 239.69

0.360 0.588 0.398 0.510 0.390 0.480 0.970

1.063 0.644 0.860 0.520 0.550 0.560 0.501 0.759 1.180 0.777

1.093 0.850 0.852 2.453 1.223 2.610 1.547 1.223 1.130 1.740 1.490 1.223 1.833 1.687 1.760 1.683 1.660 2.327 1.323

2.620 2.073 1.403 1.987 1.920 1.317 0.492 0.580 0.510 4.100 0.519 0.490 0.670 0.950

2.56 5.93 4.03 8.98 4.44 9.98 4.47 12.20 4.56 6.08 4.56 9.66 4.76

10.20

4.88 9.96 5.08 5.66 5.22 6.10 5.44 9.71 5.47 6.30 5.51 10.00 5.66 7.49 5.69 6.00 5.69 12.40 6.05

6.20 6.25 6.61 6.64 6.86 6.98 7.08 7.10 7.27 7.27 7.42 7.64 7.69 7.69 7.88 7.93 7.93 8.08 8.13

8.30 8.49 8.61 8.61 8.83 8.88 10.00 11.00 11.60 11.70 12.80 13.10 15.10 18.00

6.98

7.81 6.47 6.47 8.81 8.32 11.50 13.10 12.20 8.76 11.30 11.70 11.10 10.30 11.30 12.00 9.49 12.60 12.90 11.00

12.70 11.00 9.13 11.60 11.40 11.60 21.60 23.40 28.40 12.20 20.90 25.80 27.20 21.00



done according to final products in clothing industry. From cooperation with textile and clothing industry it is known that in departments of clothing industry general information about characteristics of fabric is given, as well as mass per unit area, thickness and number of yarns in warp and weft direction. Mentioned properties of fabrics are significant for setting up optimal technological conditions in garment production process (cutting, fusing, selection of sewing machines with optimal needle, ironing etc.). Added principal property which is very important for high quality product production is extensibility. All mentioned properties can be determined according to prescribed standard methods in laboratories of clothing industry. If all the testing has to be done in laboratories with expensive methods, production costs will be higher. Selection of mentioned properties (mass per unit area, thickness and number of yarns in warp and weft direction) for investigation and development of model for prediction extensibility are done in order to reduce gap between industry and scientific-research departments. This is very important for authors’ cooperation with textile and clothing industry because of obtained applicable results and not only for theoretical knowledge. In the investigation of ANN application for predicting of fabric tensile properties, two sets of data are used for designing of two model of ANN. The first set consists of data for 50 fabrics, and the second one of 42 fabrics. Model of ANN for the first set of data was trained, checked and tested on data for 45 fabrics and after that deviation in experimental (measuring with KES-FB1 device) and predicted (using ANN) results for all 50 fabrics are determined. Also, model of ANN for the second set of data is designed and it was trained, checked and tested on data for 39 fabrics and after that deviation in experimental and predicted results for 42 fabrics are determined. For the first model of ANN all data from Table 1 were used and for the second model of ANN first 42 data from Table 1 was used. Data in Table 1 that are marked as grey (five rows - TK556; TK091; TK317; TK110; TK417) the first model of ANN has never seen before. Data in Table 1 that are marked as bold (three rows TK482; TK258; TK050) the second model of ANN has never seen before. As input data for ANN the density of warp and weft, mass per unit area and thickness were used and as output data for ANN fabric extensibility in warp (EMT1) and weft (EMT2) direction were used. For the purpose of this work a radial basis network consisting of two layers was tested: a hidden radial basis layer with 17 or 19 neurons, and an output linear layer with 2 neurons (Figure 3). The transfer function used in the hidden layer is: f ( x ) = e –x

2

(1)

The comparison between predicted and measured data was performed after training of ANN. For the estimation of a

Radial basis network consists of two layers. The “w” is the input weight matrix, “b” is the bias vector and “F” is the transfer function. Figure 3.

neural network model, the mean squared error (MSE) was used as the performance function. This parameter measures the performance of the network according to the equation: n n (2) MSE = 1--- e2 = 1--- ( y – y )2 n

∑

i=1

i

n

∑

i, exp

i, pred

i=1

where, yi exp - experimental data, yi pred - predicted data with ANN. For a network training function trainlm (from MATLAB Neural Network Toolbox) was used, that updates weight and bias values according to Levenberg-Marquardt optimization [23]. ,

,

Results In this chapter results of ANN application in predicting of extensibility of investigated fabrics using KES-FB1 measuring device are shown. In Table 2 comparison between measured extensibility and predicted values given by ANN for all 50 fabrics are shown. The chosen network for this purpose consists of 4 neurons in the input layer, 17 neurons in the hidden layer and 2 neurons in the output layer (ANN 4-172). In Table 3 comparison between measured extensibility and predicted values given by ANN for 42 fabrics are shown. The ∆EMT1 (%) and ∆EMT2 (%) are relative errors: EMT1ANN – EMT 1 ∆EMT1 (%) = -----------------------------------------(3) ⋅ 100 % EMT 1 EMT2ANN – EMT 2 ∆EMT2 (%) = -----------------------------------------(4) ⋅ 100 % EMT 2 The chosen network for this purpose consists of 4 neurons in the input layer, 19 neurons in the hidden layer and 2 neurons in the output layer (ANN 4-19-2). All calculations were carried out with MATLAB software version 7.8.0.347 (R2009a) and Neural Network Toolbox version 6.0.2 [23]. In Figure 4 correlation between measured and predicted values of fabric extensibility in the warp direction using an ANN for all 50 fabrics are shown, and in Figure 5 in the weft direction. In Figure 6 correlation between measured and predicted values of fabric extensibility in the warp direction using an ANN for 42 fabrics are shown, and in Figure 7 in the weft direction.



Comparison of extensibility EMT, measured using KESFB1 measuring device and computed values by ANN for data of 50 fabrics EMT2ANN ∆EMT1 ∆EMT2 ANN Sign of fabric EMT1 (%) (%) (%) (%) TK 555 2.64 5.94 3.26 0.21 TK 150 4.14 9.26 2.82 3.15 TK 166 4.45 10.05 0.30 0.67 TK 524 4.51 11.46 0.79 -6.04 TK 556 4.41 10.28 -3.23 6.39 TK 523 4.33 6.22 -5.01 2.29 TK 482 4.55 10.24 -4.35 0.36 TK 076 5.05 9.81 3.47 -1.53 TK 090 5.09 5.43 0.28 -4.05 TK 466 5.20 6.06 -0.31 -0.59 TK 522 5.46 9.45 0.30 -2.64 TK 469 5.62 6.16 2.79 -2.19 TK 520 5.34 10.65 -3.01 6.49 TK 165 5.62 7.57 -0.76 1.08 TK 091 5.69 6.20 -0.07 3.27 TK 581 5.72 12.41 0.59 0.07 TK 258 5.94 7.07 -1.74 1.35 TK 111 6.13 7.86 -1.18 0.69 TK 256 6.40 6.33 2.32 -2.21 TK 257 7.02 8.73 6.26 34.99 TK 317 6.66 9.04 0.32 2.64 TK 093 6.92 8.27 0.85 -0.57 TK 316 6.99 11.33 0.11 -1.46 TK 096 8.01 11.63 13.13 -11.24 TK 049 7.17 11.14 0.96 -8.66 TK 467 7.35 10.93 1.17 -3.25 TK 100 7.26 8.74 -0.12 -0.18 TK 102 7.30 11.99 -1.60 2.50 TK 047 7.47 12.17 -2.16 9.62 TK 041 7.89 10.43 2.63 1.29 TK 110 8.45 13.11 9.86 16.02 TK 338 8.09 11.57 2.64 -3.61 TK 101 7.17 10.59 -9.60 11.63 TK 415 8.21 12.25 3.51 -2.80 TK 307 8.22 12.60 1.70 -2.31 TK 050 9.39 12.59 15.44 14.44 TK 098 8.28 12.69 -0.25 -0.05 TK 334 8.22 12.24 -3.20 11.32 TK 323 8.18 9.79 -4.96 7.22 TK 333 8.29 12.31 -3.67 6.10 TK 455 8.56 11.33 -3.01 -0.65 TK 048 8.95 11.59 0.75 -0.08 TK 259 10.64 21.13 6.38 -2.19 TK 416 10.81 23.64 -1.69 1.03 TK 417 11.66 28.34 0.54 -0.22 TK 306 11.69 12.21 -0.07 0.09 TK 260 12.16 21.39 -5.03 2.37 TK 474 13.18 25.71 0.65 -0.33 TK 473 15.15 27.09 0.34 -0.39 TK 527 17.99 21.00 -0.03 0.01 Table 2.

921

Comparison of extensibility EMT, measured using KESFB1 measuring device and computed values by ANN for data of 42 fabrics ∆EMT1 ∆EMT2 ANN EMT2ANN Sign of fabric EMT1 (%) (%) (%) (%) TK 555 2.55 5.94 -0.22 0.23 TK 150 3.99 8.95 -1.00 -0.37 TK 166 4.41 9.94 -0.68 -0.40 TK 524 4.55 12.26 1.75 0.52 TK 556 4.58 6.09 0.45 0.22 TK 523 4.56 9.67 0.08 0.11 TK 482 3.65 9.77 -23.24 -4.24 TK 076 4.90 9.96 0.38 0.04 TK 090 5.08 5.64 0.05 -0.42 TK 466 5.25 6.15 0.64 0.78 TK 522 5.48 9.73 0.68 0.18 TK 469 5.45 6.27 -0.43 -0.43 TK 520 5.46 9.98 -0.94 -0.24 TK 165 5.65 7.49 -0.19 -0.01 TK 091 5.68 6.02 -0.19 0.27 TK 581 5.69 12.42 -0.04 0.16 TK 258 5.63 5.45 -6.97 -21.98 TK 111 6.21 7.81 0.18 0.03 TK 256 6.29 6.47 0.65 0.05 TK 257 6.56 6.46 -0.68 -0.14 TK 317 6.55 8.90 -1.32 1.03 TK 093 6.85 8.31 -0.20 -0.16 TK 316 7.00 11.45 0.33 -0.39 TK 096 7.12 12.75 0.55 -2.64 TK 049 7.03 11.63 -0.93 -4.65 TK 467 7.27 8.77 0.03 0.08 TK 100 7.60 10.37 4.51 -8.21 TK 102 7.38 11.46 -0.55 -2.02 TK 047 7.67 11.58 0.39 4.29 TK 041 7.60 10.37 -1.22 0.66 TK 110 7.54 11.87 -1.95 5.03 TK 338 7.94 11.80 0.75 -1.67 TK 101 7.83 10.62 -1.31 11.95 TK 415 7.92 12.33 -0.12 -2.11 TK 307 8.18 12.93 1.19 0.21 TK 050 9.55 11.66 17.49 6.03 TK 098 8.30 12.70 -0.02 0.03 TK 334 8.58 10.63 1.01 -3.37 TK 323 8.62 9.29 0.11 1.70 TK 333 8.39 11.96 -2.56 3.13 TK 455 8.89 11.37 0.69 -0.30 TK 048 8.91 11.67 0.29 0.62 Table 3.

Discussion and Conclusion

Based on managed investigations and given results it can be discussed following. According to the data which are shown in Table 2 the maximum deviation in results of extensibility in the warp direction is determined for fabric designated as TK050 in amount of 15.44 %, and the


Correlation between measured (KES-FB1 – Targets T) and predicted (ANN 4-17-2 – Outputs Y) data for extensibility in the warp direction for all 50 fabrics. Figure 4.

Correlation between measured (KES-FB1 – Targets T) and predicted (ANN 4-17-2 – Outputs Y) data for extensibility in the weft direction for all 50 fabrics.


Correlation between measured (KES-FB1 – Targets T) and predicted (ANN 4-19-2 – Outputs Y) data for extensibility in the warp direction for chosen 42 fabrics. Figure 6.

Correlation between measured (KES-FB1 – Targets T) and predicted (ANN 4-19-2 – Outputs Y) data for extensibility in the weft direction for chosen 42 fabrics.

Figure 5.

Figure 7.

minimum deviation for fabric designated as TK527 in amount of −0.03 %. For weft direction the maximum deviation between extensibility values is determined for fabric designated as TK257 in amount of 34.99 %, and the minimum deviation for fabric designated as TK527 in amount of 0.01 %. According to data for 42 fabrics which are presented in

Table 3 the maximum deviation between extensibility values determined experimentally and predicted by an ANN for the warp direction was for fabric designated as TK482 in amount of −23.24 %, and the minimum deviation for fabric designated as TK467 in amount of 0.03 %. For weft direction the maximum deviation between extensibility values is determined for fabric designated as TK258 in amount of


−21.98 %, and the minimum deviation for fabric designated as TK165 in amount of −0.01 %.

Coefficient of correlations between experimentally data from KES-FB1 measuring device and predicted data given by an ANN were determined for both direction (warp and weft) and for data values for 50 and 42 fabrics. Coefficient of correlations determined for data set for 50 fabrics were higher. For warp direction coefficient of correlation determined on data for 50 fabrics was r=0.9934 (Figure 4), and for weft direction r=0.9930 (Figure 5). For data set of 42 fabrics coefficient of correlation for warp direction was r=0.9840 (Figure 6), and for weft direction r=0.9852 (Figure 7). Based on managed investigations and given results it can be concluded that an artificial neural network can be used for prediction of fabric properties. Between data which are given experimentally and data predicted using an ANN high coefficient of correlation are determined (r≥0.9840). If one wishes to predict extensibility for fabrics with different parameters (for example with different weave or raw material composition) then another set of fabric parameters will be used for training, checking and testing and new ANN model is needed. The paper is made as a step in connecting scientific work and economy what is nowadays fundamental for managing investigation and development in science and economy. With application of this investigation we can contributed to development of model for predicting different properties of fabrics. With the proposed model it is possible to decrease number of experimental measuring which is often unavailable to garment industry because described measuring system is very expensive and development and investigated division in industry can not afford it. Determination of other fabric parameters as density of warp and weft, thickness and mass per unit area are available in textile industry, so this approach for modelling of fabric properties is applicable for them. This paper is related to fabric extensibility because this is one of the most important fabric parameters and which influence on ergonomic comfort of garment. The further work will be continued with investigation of other parameters of fabrics, which are important in investigations studies of problems, which appear during garment manufacturing processes, and also in investigation of the influence of fabric properties on final appearance of different garments.

Acknowledgements The research presented in this paper is a part of the research project “Computational Modelling in Engineering Analyses of Textiles and Garment” which is financially supported by the Ministry of Science Education and Sport of


923

the Republic of Croatia under contract no. 117-11718791899.

References 1. www.optitex.com. 2. www.lectra.com. 3. P. Volino, F. Cordier, and N. Magnenat-Thalmann, Comput. Aided Design, 37, 593 (2005). 4. S. Liu, J. C. Sheng, L. He, and F. Chen, “Proceeding of the International Conference on Computer Science and Software Engineering”, Vol. 2, pp.1082-1085, 2008. 5. F. T. Peirce, J. Text. Inst., 16, T45 (1937). 6. A. De Boos and D. Tester, CSIRO Report No. WT92.02, 1994. 7. P. G. Minazio, Int. J. Cloth. Sci. Tech., 7, 43 (1995). 8. S. Kawabata, M. Niwa, and F. Wang, Text. Res. J., 64, 597 (1994). 9. G. K. Stylios, “Proceeding of the 2nd International Textile, Clothing & Design Conference”, Vol. 1, pp.3-12, Croatia, 2004. 10. D. Zavec Pavliniéc, J. Geršak, J. Demšar, and I. Bratko, Text. Res. J., 76, 235 (2006). 11. D. Zavec Pavliniéc, and J. Geršak, Int. J. Cloth. Sci. Tech., 16, 252 (2004). 12. J. Lindberg, L. Westerberg, and R. Svennson, J. Text. Inst., 51, 1475 (1960). 13. H. Morooka and M. Niwa, J. Text. Machin. Soc. Japan, 24, 105 (1978). 14. J. Geršak, Tekstil, 52, 368 (2003). 15. D. Zavec Pavliniéc, Ph.D. Dissertation, Faculty of Mechanical Engineering University of Maribor, Maribor, 2005. 16. S. Kawabata, “The Standardization and Analysis of Hand Evaluation”, 2nd ed., pp.28-70, The Hand Evaluation and Standardization Committee, Osaka, 1980. 17. S. Haykin, “Neural Networks, A Comprehensive Foundation”, 1st ed., pp.35-82, Prentice Hall, USA, 1994. 18. L. Sang-Song, S. Tien-Wei, and L. Jer-Yan, J. Text. Eng., 48, 43 (2002). 19. R. Beltran, L. Wang, and X. Wang, Text. Res. J., 74, 757 (2004). 20. V. Bevilacqua, L. Cariello, G. Mastronardi, V. Palmieri, and M. Giannini, “Proc. of the ICIC 2008”, Vol. 1, pp. 1166-1174, 2008. 21. P. C. L. Hui, K. C. C. Chan, K. W. Yeung, and F. S. F. Ng, Int. J. Cloth. Sci. Tech., 19, 291 (2007). 22 D. Golob, Ð. Parac Osterman, and J. Zupan, Fibres Text. East. Eur., 16, 93 (2008). 23. The MathWorks - MATLAB and Simulink for Technical Computing, Available: http://www.mathworks.com/.