Comparison between KES-FB and FAST in discrimination of fabric characteristics

Abstract

In order to provide a basis for parameter setting during the process of making man-made fiber fabrics as imitations of natural fibers fabrics, to understand characteristics of natural fiber fabrics and to establish discriminant methods for such characteristics have become important subjects in the textile industry nowadays. By applying discriminant analysis and neural network, I have successfully used fabric physical properties measured by KES-FB system and FAST system to establish discriminant models for fabric characteristics of cotton, linen, wool and silk. KES-FB discriminant model that applied fabric mechanical properties appeared to have a better classification ability than FAST discriminant model. On the other hand, the discriminant model established by applying neural network appeared to have a better classification ability than that by applying discriminant analysis. While only 9 and 5 physical properties of fabrics measured by KES-FB system and FAST system, respectively, needed to be applied, the discriminant model established by applying stepwise method should possess the characteristics of simplicity, convenience and effectiveness.