Design and Selection of Parameters and Training Method of Advanced BP Network

This paper Proposes a mean to design the improved BP network based on self-adaptive learning rate and additional momentum, through detailed design and process steps, discussed influence and relationship of the amount of hidden layer neurons, initial weights, learning rate and other parameters in the network design process, analyzed the announcements in training algorithm, conducted the corresponding training and testing of network, obtained a neural network model of simple calculation, fast convergence, high precision error.