Predicting compression and surfaces properties of knits using fuzzy logic and neural networks techniques

Purpose - The purpose of this paper is to model the relationship between manufacturing parameters, especially finishing treatments and instrumental tactile properties measured by Kawabata evaluation system. Design/methodology/approach - Two soft computing approaches, namely artificial neural network (ANN) and fuzzy inference system (FIS), have been applied to predict the compression and surface properties of knitted fabrics from finishing process. The prediction accuracy of these models was evaluated using both the root mean square error and mean relative percent error. Findings - The results revealed the model's ability to predict instrumental tactile parameters based on the finishing treatments. The comparison of the prediction performances of both techniques showed that fuzzy models are slightly more powerful than neural models. Originality/value - This study provides contribution in industrial products engineering, with minimal number of experiments and short cycles of product design. In fact, models based on intelligent techniques, namely FIS and ANNs, were developed for predicting instrumental tactile characteristics in reference to finishing treatments.