PREDICTING THE PHYSICAL PROPERTIES OF DRAWN NYLON-6 FIBERS USING AN ARTIFICIAL-NEURAL-NETWORK MODEL

Low-oriented nylon-6 fibers were drawn in a multistage drawing process, during which the number of drawing steps and the temperature of each step were changed. The physical properties of these fibers were measured and compared with the values predicted by a multiple-linear-regression model. Moreover, six input variables and four output variables were used in an artificial neural network (ANN) to establish the logical relationships between the inputs and outputs. Attempts were also made to determine the effective parameters for each physical property and explain the observed trends. The results showed that the models based on the ANN performed well and provided stable responses in predicting combined interactions between independent variables.