Optimization of Plasma Fabric Surface Treatment by Modeling with Neural Networks
Artificial neural networks (ANNs) are used to model the relationship between plasma processing parameters and woven fabric surface wetting properties. In this model, fourteen features characterizing woven structures and two plasma parameters are taken as input variables and the water contact angle cosine and capillarity height as output variables. In order to reduce the complexity of the model, a fuzzy logic–based method is used to select the most relevant parameters that are taken as inputs of the reduced neural model. Two techniques (early stopping and Bayesian regularization) are used for improving the generalization ability of neural networks. A methodology for optimizing such models is described. Moreover, a connection weight method is used to investigate the relative importance of each input variable. From the experiments, we find a good agreement between experimental and predicted data and that Bayesian regularization technique is the most suitable to achieve a good generalization.
KeywordsArtificial neural networks Fuzzy logic–based selection criterion Industrial modeling Atmospheric air plasma Woven fabrics Wettability
- 6.Yousefi MMR, Mirmomeni M, Lucas C (2007) Input variables selection using mutual information for neuro fuzzy modeling with the application to time series forecasting. In: Proceedings of international joint conference on neural networks, Orlando, Florida, USA, pp 1121–1126Google Scholar
- 10.AbdJelil R, Zeng X, Koehl L, Perwuelz A (2009) Neural model of woven fabric surface modification by atmospheric air plasma. In: Intelligent textiles and mass customization international conference, Casablanca, MoroccoGoogle Scholar
- 12.Gil P, Cardoso A, Palma L (2009) Estimating the number of hidden neurons in recurrent neural networks for nonlinear system identification. In: IEEE international symposium on industrial electronics. IEEE Press, Seoul, South Korea, pp 2053–2058Google Scholar
- 13.Lee KW, Lam HN (1995) Optimal sizing of feed forward neural networks: case studies. In: 2nd New Zealand two-stream international conference on artificial neural networks and expert systems. IEEE computer society, Dunedin, New Zealand, pp 79–82Google Scholar