Investigation into the curling behavior of single jersey weft-knitted fabrics and its prediction using neural network model

This research investigates the effect of fiber, yarn, and fabric parameters on curling phenomenon of single jersey weft-knitted fabrics which is interpreted to have curling surface in both course and wale direction. Taguchi’s experimental design is used to estimate the optimum process conditions and to examine the individual effects of all controllable factors on curling one by one. The controllable factors are blending ratio of polyester to cotton fiber, yarn twist and count, fabric structure, knit density, and relaxation time. Results show that fabric structure and knit density have the most dominant effect on the fabric curling. The optimum conditions of minimum curling values were also determined. Finally, the curling surface in course and wale direction as a two features of curling phenomenon was predicted using artificial neural network which selects scale conjugate gradient learning algorithm based on process parameters of single jersey weft-knitted fabrics. Our findings confirm the good capability of artificial neural network algorithm to predict these features.