Have a personal or library account? Click to login
Prediction of Mechanical Properties of Woven Fabrics by ANN Cover

Prediction of Mechanical Properties of Woven Fabrics by ANN

Open Access
|Dec 2022

Figures & Tables

Fig. 1

Neural Network architecture for all mechanical properties
Neural Network architecture for all mechanical properties

Fig. 2

Overall Training Performance of ANN model
Overall Training Performance of ANN model

Comparison between actual and predicted values of properties tested in the weft direction

RunTensile strength (N)“Stiffness”Bending length * 10−2 (m)Elongation %
ActualPredictedActualPredictedActualPredicted
1258.00258.002.102.1517.7017.60
2234.00234.002.001.9513.7013.50
3321.00321.002.402.3721.8021.20
4307.00307.002.202.2021.2021.20
5251.00251.002.102.1113.5013.30
6290.00290.002.102.0920.5020.50
7241.00241.002.002.0115.1014.60
8312.00312.002.302.3120.7021.10
9290.00290.002.102.0920.5020.50
10228.00228.002.001.9913.1013.50
11284.00284.002.002.0019.2019.20
12207.00207.001.901.9213.2013.40
13290.00290.002.202.2221.7021.60
14284.00284.002.002.0019.2019.20
15290.00290.002.202.2221.7021.60

Factors and levels of weft yarns

Levels
Factors123
X1Weft density (picks/m)0.230.250.27
X2Weft yarn count (Nm)40/150/1------
X3Fiber blend ratio of weft yarn Polyester (PE %)0%50%65%
X4Fiber blend ratio of weft yarn Cotton (C%)100%50%35%

Experimental model

RunX1X2X3X4
Picks/mWeft yarnPE %C %
count (Nm)
11111
21122
31133
41211
51222
61233
72111
82122
92133
102211
112222
122233
133111
143122
153133
163211
173222
183233

Comparison between the prediction performance of all properties by ANNs

Tested propertiesTensile strength (N)“Stiffness” Bending length * 10−2 (m)Elongation %
Statistical factorswarpweftwarpweftwarpweft
R-squared :coefficient of determination1.001.000.970.980.990.99
MSE :mean squared error0.000.000.000.000.040.07
RMSE :root mean squared error0.040.030.050.020.200.27
MAE :mean absolute error0.030.020.040.020.150.19
MAPE: mean absolute percentage error0.01%0.01%1.62%0.88%0.69%1.14%

Comparison between actual and predicted values of properties tested in the warp direction

RunTensile strength (N)“Stiffness” Bending length * 10−2 (m)Elongation %
ActualPredictedActualPredictedActualPredicted
1357.00357.002.302.3920.5020.90
2334.00334.002.202.1519.4019.30
3377.00377.002.702.6722.4022.50
4368.00368.002.502.5022.3022.00
5358.00358.002.402.4222.0022.00
6360.00360.002.402.3522.6022.60
7368.00368.002.001.9520.7020.70
8369.00369.002.502.5323.1023.40
9360.00360.002.402.3522.6022.60
10351.00351.002.202.1921.6021.70
11350.00350.002.302.3221.3021.40
12331.00331.002.002.0915.7015.70
13353.00353.002.402.4223.3023.00
14350.00350.002.302.3221.3021.40
15353.00353.002.402.4223.3023.00

ANN Training Algorithm

-Train network using Levenberg-Maquardt back-propagation
- Activation function: (trainlm).-Hidden layer size =10
-Performance: Mean squared error (mse)-Gradient: 1.00e-05
DOI: https://doi.org/10.2478/ftee-2022-0036 | Journal eISSN: 2300-7354 | Journal ISSN: 1230-3666
Language: English
Page range: 54 - 59
Published on: Dec 11, 2022
Published by: Łukasiewicz Research Network, Institute of Biopolymers and Chemical Fibres
In partnership with: Paradigm Publishing Services
Publication frequency: Volume open

© 2022 Sherien N. Elkateb, published by Łukasiewicz Research Network, Institute of Biopolymers and Chemical Fibres
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.