Prediction of Properties of Unknotted Spliced Ends of Yarns Using Multiple Regression and Artificial Neural Networks. Part I: Identification of Spliced Joints of Combed Wool Yarn by Artificial Neural Networks and Multiple Regression
Research and development
Authors:
- Lewandowski Stanisław
Institute of Textile Engineering and Polymer Materials, University of Bielsko-Biala, Bielsko-Biała, Poland - Drobina Robert (j/w)
 Full text | Abstract: Applying the software environment Statistica for neural networks allowed the use of artificial neural networks and regression analysis to predict the physical properties of unknotted joints of yarn ends. The database entered into the network was built on the basis of determining characteristic geometric dimensions and the strength properties of joints, as well as assessing non-additive features, represented by teaseling and tangling. Networks of the multilayer perceptron type (MLP) and generalized regression neural networks (GRNN) were used. In order to compare the results, multiple regression was also applied. |
Tags: combed wool yarn, pneumatically spliced joints, additive quantities, non-additive features, artificial neural network, multilayer perceptron, generalized neural network, back propagation algorithm.