Measuring The Degree Of Twist Set Based On The Twist Intrinsic Definition And Its Prediction By Artificial Neural Network

Twist fixation is an important part of yarn production in two methods of virtual or real twisting method. Currently, twist fixation is calculated by performing heat treatment based on a combination of stabilization of bending, strain, and torque according to common methods. This is the fact that the basis of yarn twist is simply the application of twisting force on the fibres, and as a rule, twist stabilization should be the same twisting force. In the new method, a way to evaluate the degree of stabilization of the twist, which is only caused by the torsional force applied to the fibres, is presented. With the help of Taguchi method, 25 samples were prepared under different levels of twist, temperature and time. The design is done for three factors (time, temperature, twist) and five levels for each factor. The twist, temperature and time levels are respectively (0,40,80,120,180), (23,60,90,120,150) and (0,60,90,120,150). Then heat treatment was done and the degree of twist stabilization was calculated according to the new method and predicted by regression and neural network. The results show a good correlation between the experimental and predicted results by the ANN model compared to the regression method.