Predicting bending rigidity of woven fabrics by neuro-genetic hybrid modeling

Abstract

The possibility of prediction of bending rigidity of cotton woven fabrics with the application of Neuro-genetic model has been explored. For this purpose, number of cotton grey fabrics meant for apparel end-use was desized, scoured, and relaxed. The fabrics were then conditioned and tested for bending properties. A feed-forward neural network model was first formed and trained with adaptive learning rate back-propagation with momentum. In the second model, a hybrid learning strategy was adopted. A genetic algorithm was first used as a learning algorithm to optimize the number of neurons and connection weights of the neural network. Later, a back-propagation was applied as a local search algorithm to achieve global optima. Results of hybrid neural network model were compared with that of back-propagation neural network model in terms of their prediction performance. Results show that the prediction by Neuro-genetic model is better in comparison with that of back-propagation neural model.