The main objective of this research is to predict the mechanical properties of viscose/lycra plain knitted fabrics by using fuzzy expert system. In this study, a fuzzy prediction model has been built based on knitting stitch length, yarn count, and yarn tenacity as input variables and fabric mechanical properties specially bursting strength as an output variable. The factors affecting the bursting strength of viscose knitted fabrics are very nonlinear. Hence, it is very challenging for scientists and engineers to create an exact model efficiently by mathematical or statistical model. Alternatively, developing a prediction model via ANN and ANFIS techniques is also difficult and time consuming process due to a large volume of trial data. In this context, fuzzy expert system (FES) is the promising modeling tool in a quality modeling as FES can map effectively in nonlinear domain with minimum experimental data. The model derived in the present study has been validated by experimental data. The mean absolute error and coefficient of determination between the actual bursting strength and that predicted by the fuzzy model were found to be 2.60% and 0.961, respectively. The results showed that the developed fuzzy model can be applied effectively for the prediction of fabric mechanical properties.
Knitting is one of the main fabric manufacturing methods among the knitting, weaving, and nonweaving in the textile manufacturing. Basically, knit fabric is formed by intermeshing yarn loops with each other in wale and course directions. The quality of fabrics is considered a big issue in the global textile and apparel market. The demand of knitted fabric especially viscose knitted fabric is increasing rapidly due to their unique quality characteristics such as elasticity, drape, wrinkle resistance, comfort, softness, and easy-care properties over woven fabrics. Viscose knitted fabrics are very popular for apparel wears such as T-shirts, shirts, sweaters, blouses, underwear, casual wear, active wear, and sportswear because of their distinctive quality characteristics as compared to woven fabrics [
The bursting strength, however, is one of the most important mechanical properties among all the viscose plain knitted fabrics qualities. Knitted fabrics are not just rendering forces in the vertical and perpendicular directions but also they are exposed to multiaxial forces during dyeing, finishing, and usage. Testing tensile and tearing strength in the wale and coarse directions in knitted fabrics are not suitable because of the structural properties; hence, testing bursting strength turns out to be extremely important particularly for knitted fabrics before manufacturing. Generally, the bursting strength test is conducted to evaluate the fabric’s capability to withstand multiaxial stresses without breaking off [
Basically, strength of fabrics depends on starting fabric forming materials such as yarn properties and fabric density. The literature review revealed that various studies have been reported on the factors affecting the bursting strength of knitted fabric including yarn type, yarn count, yarn tenacity, yarn breaking elongation, yarn breaking strength, yarn twist, yarn evenness, fabrics GSM, fabric wale and courses, knitting stitch length, cover factor, tightness factors, and relaxation treatment [
Moreover, all these factors affecting the bursting strength of knitted fabrics are nonlinear and interactive with other. Hence, it poses a great challenge for scientists and engineers to develop effectively an exact model based on mathematical and statistical techniques [
On the other hand, the intelligent systems such as artificial neural network (ANN) and adaptive neuroinference system (ANFIS) models which have the ability to model in nonlinear domain have been applied by few researchers in the previous research for predicting the bursting strength of knitted fabrics. Jamshaid et al. applied ANFIS model to predict the bursting strength of plain knitted fabrics as a function of yarn tenacity, knitting stitch length, and fabric GSM [
In this regard, fuzzy expert system (FES) is found to be the scientific and engineering solution for quality modeling as FES provides reliable prediction outcome with small experimental data in nonlinear and ill-defined textile domain [
The key purpose of this work is to construct fuzzy intelligent model for predicting the bursting strength of viscose/lycra plain knitted fabrics as a function of knitting stitch length, yarn count, and yarn tenacity, which is not reported in the published literature.
The artificial intelligence fuzzy expert system is a structure of multivalued logic and an extension of crisp logic derived from fuzzy mathematical set theory developed by Zadeh in 1965 [
Fundamental unit of a fuzzy expert system [
In (
As an expression, when a fuzzy model with two inputs and one output is involved, then development of fuzzy inference rules can be presented as follows.
where
The prediction accuracy of the developed model has been investigated according to the global prediction error such as mean absolute error (MAE) and coefficient of determination (
The coefficient of determinations (
For development of fuzzy prediction model, three knitting variables such as knitting stitch length (SL), yarn count (YC), and yarn tenacity (YT) have been used as input variables and bursting strength of knitted fabrics as output variable. These knitting variables have been exclusively selected as they influence the fabric bursting strength considerably. A fuzzy logic Toolbox from MATLAB (version 7.10.0) was used to develop the proposed fuzzy model of bursting strength. The construction of fuzzy modeling for bursting strength has been depicted in Figure
Schematic diagram of fuzzy modeling for bursting strength [
For fuzzification, four possible linguistic subsets, namely, very low (VL), low (L), medium (M), and high (H), for input variables SL and YC and three convenient linguistic subsets, namely, low (L), medium (M), and high (H), for input variable YT were chosen in such a way that they are evenly spaced and cover up the entire input spaces. Ten output fuzzy sets (Levels 1 to 10) (where
The triangular shaped membership functions for the fuzzy variables, namely, stitch length (SL), yarn count (YC) and yarn tenacity (YT), and bursting strength (BS), have been created using fuzzy logic Toolbox from MATLAB software (version 7.10.0) and are shown in Figures
Membership function of input variable “SL.”
Membership function of input variable “YC.”
Membership function of input variable “YT.”
Membership function of output variable “BS.”
The coefficients of membership functions for the fuzzy inference system (FIS) parameters are shown in Table
Coefficients of membership functions for FIS parameter of BS.
Linguistic variables | Type | Coefficients (%) | ||
---|---|---|---|---|
|
|
| ||
Level 1 | Z-shaped | 250 | 270 | — |
Level 2 | Triangular | 250 | 270 | 290 |
Level 3 | Triangular | 270 | 290 | 310 |
Level 4 | Triangular | 290 | 310 | 330 |
|
Triangular | 310 |
|
350 |
Level 6 | Triangular | 330 | 350 | 370 |
Level 7 | Triangular | 350 | 370 | 390 |
Level 8 | Triangular | 370 | 390 | 410 |
Level 9 | Triangular | 390 | 410 | 435 |
Level 10 | Triangular | 410 | 435 | 460 |
Conceptually, there could be 4 × 4 × 3 = 48 fuzzy rules, as input variable SL having 4 linguistic levels, YC having 4 linguistic levels, and YT having 3 linguistic levels. However, in order to make the fuzzy expert system simpler, only 24 fuzzy rules have been formed based on expert knowledge and previous experience [
Fuzzy inference rules.
Rules number | Input variables | Output variable | ||
---|---|---|---|---|
SL | YC | YT | BS | |
1 | VL | H | L | Level 3 |
2 | L | H | L | Level 2 |
⋮ |
|
|
| |
10 | L | L | M | Level 5 |
11 | M | L | M | Level 6 |
|
|
|
| |
23 | M | VL | M | Level 9 |
24 | H | M | H | Level 6 |
To demonstrate the fuzzification process, linguistic expressions and membership functions of stitch length (SL), yarn count (YC), and yarn tenacity (YT) obtained from the developed rules and above formula (
In this research, total 12 viscose/lycra blended plain fabrics samples were knitted according to Table
Knitted fabric variables and their level.
Process parameters | Unit | Level | |||
---|---|---|---|---|---|
1 | 2 | 3 | 4 | ||
Stitch length | mm | 2.7 | 2.8 | 2.9 | 3.0 |
Yarn count | Ne | 24 | 30 | 34 | |
Yarn tenacity | g/tex | 15.25 | 15.5 | 15.75 |
Circular knitting machine (APS Textile, Bangladesh).
The fabrics samples were subjected to semibleached at 90°C for 40 min in a sample dyeing machine using anticreasing agent (Kappavon CL 1 g/L), sequestering agent (Sirrix 2UD 0.5 g/L), wetting agent (Felosan NOF 1 g/L), soda ash (2.5 g/L), hydrogen peroxide 50% (1 g/L), and stabilizing agent (0.2 g/L). Then the fabrics samples were washed with proper rinsing and finally treated with acetic acid (1 g/L) and peroxide killing agent (0.2 g/L) for neutralizing and peroxide killing, respectively. After bleaching, the fabric samples were dried in an open stenter and compacted properly. After production, all the fabrics samples were conditioned on a flat surface first for at least 24 hours prior to testing under standard atmospheric conditions at relative humidity (
The graphical operation of the fuzzy prediction model has been depicted with an example in Figure
Rule viewer.
Surface plot showing the impact of stitch length and yarn count on bursting strength.
Surface plot showing the impact of stitch length and yarn tenacity on bursting strength.
Effects of stitch length, yarn count, and yarn tenacity on bursting strength have been shown in Figures
Effect of stitch length and yarn count on bursting strength.
Effect of stitch length and yarn tenacity on bursting strength.
A similar phenomenon has been observed for yarn count on bursting strength as shown in Figure
In contrast, the effect of yarn tenacity on the bursting strength is much more reflective as compared to stitch length as shown in Figure
From this investigation, it is clearly observed that yarn tenacity has the greatest and main effect on bursting strength as compared to knitting stitch length and yarn count. Therefore, it is very important to maintain optimum level of knitting parameter in the knitting process in order to achieve the required bursting strength with good quality fabrics.
The developed fuzzy prediction model has been validated by experimental data. Prediction was done using the fuzzy logic rule viewer. The results from the developed fuzzy model were then compared with 12 validated experimental results as shown in Table
Comparison of predicted and experimental values of fabric bursting strength.
Number | Stitch length (mm) | Yarn count (Ne) | Yarn tenacity (g/tex) | Predicted bursting strength | Experimental bursting strength | Absolute error (%) |
---|---|---|---|---|---|---|
1 | 2.7 | 34 | 15.25 | 310 | 308 | 0.65 |
2 | 2.8 | 34 | 15.25 | 290 | 290 | 0.00 |
3 | 2.9 | 34 | 15.25 | 310 | 324 | 4.32 |
4 | 3 | 34 | 15.25 | 278 | 267 | 4.12 |
5 | 2.7 | 30 | 15.5 | 350 | 351 | 0.28 |
6 | 2.8 | 30 | 15.5 | 330 | 342 | 3.51 |
7 | 2.9 | 30 | 15.5 | 350 | 329 | 6.38 |
8 | 3 | 30 | 15.5 | 324 | 310 | 4.52 |
9 | 2.7 | 24 | 15.75 | 452 | 468 | 3.42 |
10 | 2.8 | 24 | 15.75 | 410 | 403 | 1.74 |
11 | 2.9 | 24 | 15.75 | 452 | 458 | 1.31 |
12 | 3 | 24 | 15.75 | 410 | 406 | 0.99 |
Mean absolute error (%) |
|
|||||
Coefficient of determination ( |
|
Correlation between experimental and fuzzy model predicted values of bursting strength.
In this research investigation, fuzzy model has been developed for predicting the fabric mechanical properties like bursting strength of viscose/lycra plain knitted fabric. The prediction model was made by taking the knitting stitch length, yarn count, and yarn tenacity as input variables and fabric bursting strength as output variable. The developed prediction model confers an excellent understanding about the interaction between knitting process variables and their effects on the fabric bursting strength. From the experimental study, it has been found that yarn tenacity has the greatest and main effects on the fabric bursting strength than that of yarn count and knitting stitch length. The fuzzy model derived in this research has been verified from the experiment data.
The mean absolute error (MAE) and coefficient of determination (
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors are thankful to the management of textile research laboratory of APS and APS Textile, APS Group Bangladesh for providing the facilities for this research work.