Fig. 8

Performance comparison between one-hidden-layer (a) and two-hidden-layer (b) neural networks. FFBP neural network structure is used in both a and b, the used training function is trainbr. In a, logsig function is used as transfer function to train the 85 neurons. In b, we have two layers of neurons; the first layer consists of 45 neuron with tansig transfer function, while the second layer consists of 25 neurons with logsig transfer function