Prediction of Filling Power of Expanded Cut Tobacco with Neural Network

In order to find the relationship between the composition of tobacco leaf and the filling power of expanded cut tobacco,the quantitative relation between them was preliminarily modeled under steady processing parameters by feed-forward back-propagation artificial neural network.The powerful ability of nonlinear function approximation of neural network was demonstrated by choosing suitable transfer function.After training with abundant data,the model possessed good prediction ability and the relative error of predictive results was around 5%.