PREDICTION OF PLASMA SURFACE MODIFICATION OF WOVEN FABRICS USING NEURAL NETWORKS

In this paper, artificial neural networks are used to investigate the relationship between plasma processing parameters and woven surface wetting properties. In order to reduce the model complexity, a fuzzy criterion is used to select the most relevant parameters which are taken as inputs of the ANNs. The outputs are the surface water contact angle and the capillarity of woven fabrics. The use of early stopping and Bayesian regularization approaches are considered. Two different network configurations are studied. One deals with two networks having each one output layer and another with a single network combining the two outputs. Obtained results show that the first configuration combined with the Bayesian regularization approach is the most suitable to achieve a good generalization..