Predicting the Hairiness of Cotton Rotor Spinning Yarns by Artificial Intelligence

Hairiness is one of the most important parameters affecting fabric quality in textile industries. Up to now, many researchers have focused their studies on the hairiness and its related concepts. It is well known that fiber properties affect hairiness, nevertheless, spinning machine parameters which are also effective on hairiness are not well studied before. In this study, the prediction ability of hairiness by taking account of the variables including rotor type, rotor diameter, doffing-tube nozzle and torque-stop was studied using support vector machine (SVM) and adaptive neuro fuzzy interface system models. Moreover, the genetic algorithm was applied to ensure that the model parameters were optimized correctly. Then, the obtained results were compared with those provided by artificial neural network (ANN) and it was revealed that all models had the great potential to be used in hairiness prediction (mean absolute percentage error = 3.8-3.9). The performances of SVM and ANN models were almost the same, however, they were better than that of fuzzy model.