Prediction of Random-Velour Needle Punching Force Using Artificial Neural Network

Random-velour needling technology is a modified version of conventional needling process. Properties of the random-velour needled fabric are controlled by the structural alteration that occurs during random-velour needling, is due to re-orientation of fibers within the pre-consolidated fibrous assembly by special fork needles. This interaction results in creation of a dual structure, comprising base and pile layers. In this work the effect of needling parameters and fiber characteristics on force exerted on the fork needle was investigated. The effect of principal parameters on total average force "Frms" exerted on individual fork needle was determined using an Artificial Neural Network (ANN) modeling and the error percentage of absolute average of predicted tests data was also calculated. Significance percentages of input parameters on "Frms" was indicative of the similar influence of fiber characteristics and needling parameters on "Frms". Results manifested the importance of punch density and barbed needle penetration depth during initial consolidating needle-felting operation. Result of the neural network assessment testified that the network in general was capable of mapping input and output parameters.