Optimizing drawing frame variables to enhance polyester spun yarn quality using soft computing techniques

Abstract

This study explores optimizing the production of 100% polyester spun yarn by analyzing key drawing frame variables, such as break draft, total draft, back roller gauge, and front roller gauge, using a systematic and algorithmic approach. Response Surface Methodology (RSM) was employed to design the experiment and produce yarn samples. The yarn mass variation per unit length (CVm%) and the Imperfection Index (IPI), which includes the total count of thick places, thin places, and neps, were modeled using an Artificial Neural Network (ANN) and optimized with a Genetic Algorithm (GA). These two yarn quality responses significantly determine the appearance quality of textiles. Sensitivity analysis of the ANN models revealed that break draft was the most influential input variable, contributing 35.58% and 28.72% to the models for CVm% and IPI, respectively. The optimal drawing frame variables were determined as a break draft of 1.33, a total draft of 7, a back roller gauge of 60.5 mm, and a front roller gauge of 47.5 mm, yielding CVm% and IPI values of 11.91% and 22.82, respectively. This study integrates RSM, ANN, and GA to present a innovative approach for optimizing yarn spinning. The findings not only enhance yarn quality with greater precision and efficiency but also offer valuable insights applicable to broader textile manufacturing processes.