Application of Neural Network to Predict the Properties of Air-jet Spun Yarns

摘要

The paper deals with the predictive modelling of low-stress mechanical and surface properties of air-jet yarns spun from 100% polyester fibres. The yarn properties like flexural rigidity, compressional energy and hairiness were predicted for different levels of process variables in air-jet spinning using artificial neural network. An inverse model of the neural network to predict process variables in air-jet spinning that will yield a given set of yam properties has also been proposed. The inverse model could be expected to play a vital role in controlling, modelling and predicting spinning process.