Prediction of dimensional change in finished fabric through artificial neural networks

Abstract

When anti-shrinkage precaution is taken for finishing processes, shrinkage could be observed with cotton and viscose fabrics by 8-15% and 20%, respectively. Therefore, capability of estimation of shrinkage rate for fabrics at the end of finishing would be a significant advantage. This study tried to estimate the shrinkage of single jersey and interlock fabrics at the end of relaxation processes by means of the Artificial Neural Networks (ANN). To that end totally 72 varieties of fabric were manufactured in two groups of the elastane and the non-elastane fabrics. Then, in each of two groups included 36 different varieties on the basis of single jersey and interlock weaving types using six different raw materials in three different densities. The processes were applied to fabrics during finishing process are thermo-fixing, washing, drying and sanforizing process. ANN model was used to predict dimensional change at the end of the sanforizing. For ANN, the two-layer feed-forward perceptron, also called single hidden layer feed-forward neural network was used to estimate dimensional change of width and length. Finally, the ANN exhibited successful performance in prediction of dimensional change in fabrics. The prediction of the dimensional properties produced by the neural network model was proved to be highly reliable (R2> 0.98). © 2018 Ege Universitesi. All rights reserved.