Predication on Pipeline Transportation Expense based on Artificial Neural Networ k
-
-
Abstract
Pipeline transportation expense predicting model is set up based on artificial n eural network, for artifi-cial neural network can simulate the relation between result variable and uncert ainty cause variable. The model's network structure has input-layer (one node), latent-layer (seven nodes) and output-layer (one node). The 25 study samples' training is completed using the improved BP algorit hm and each node's weight and bias value are gained. Then given transport quantity pipeline transpo rtation expense is predicted by using trained network model. The 6 sample's predicting result indicates that the error of predicting value range is within 4% using this model. It completely satisfies engineering p ractice needs.
-
-