Prediction Of Shrinkage For Circular Knitted Fabric Using Regression Analysis And Artificial Neural Networks (ANNs) <h6 class="resdate">(6/7/2010 18:00)</h6>

Shrinkage is a great problem of circular knitted fabric. Where if can not control it, you will loss the fabric quality and the consumer , so the aim of this paper how can you predict it. Starfish is short for “start as you mean to finish it ’’ is a computer program and a bay of know-how which was applied to demonstrate how to engineer circular cotton knits so the quality and performance can be right first time and on time .In our work prediction of fabric shrinkage and modeling knitting quality were determined by applying starfish procedure with the help of Regression Analysis & ANNs in order to save time of planning and obtain precise adjustments of machine in few minutes and saving cost by saving consumption of material required to make trial and error .Experiments were produced on Interlock (1*1) and four designs of Pique fabrics by varying count &loop length. The row knitted fabrics were dyed and finished trough different systems. Percentage lengthwise and widthwise shrinkage were determined for every working stage. Prediction has been done by traditional Regression Analysis and two techniques of ANNs [Easy Neural Network(ENN) and AI-Net(AIN) programs]. Regression failed to give accurate predicted values for both fabrics in some cases. The percentage error range for all applied techniques as following:
for regression analysis (-23% – 14% ),while ENN (1.4% – 31.2% ) and AIN( -12% – 7% ) for Interlock. But for Pique regression analysis ( -31.2% – 23% ),while ENN ( 0.46% – 14.7% ) and AIN( -21% – 25% ) .The results obtained from both types of the ANNs proved that this technique can be successively applied for predicting shrinkage for both Interlock and pique fabrics.

By: El Kateb Sh. Graduated Student, Textile Engineering Dept., Faculty of Engineering, Alexandria Unive

Submit Date: 6/7/2010 18:00

close
Start typing to see posts you are looking for.
Sidebar Scroll To Top