Modelling of tensile properties of needle-punched nonwovens using artificial neural networks

The mode~ling of tensile properties of needle-punched nonwoven fabrics produced from the blends of jute and polypropylene fibres with varying fabric weight, needling density and blend ratio has been done. The tenacity and initial modulus values of needle-punched nonwoven fabrics have bee n predicted with the help of empirical model (using multiple regressIOn analysIs) and artIficIa l neural networks and compared with the experimental val ues. The artifici al neural network model is found to be much better and more accurate than an empirical model. An attempt has als o been made for the experimental verificat ion of the predicted valu es for extrapol ated input variables. The prediction by artificial neural network model sho ws better results th an that by empirical model even fo r the extrapolated input variab les.