Research on state differential artificial neural network

In this paper, an emerging artificial neural network is proposed and researched. The differential of exciting intensity of each neuron is mutually feedback to each other in the network. Hence the overall network turns out to be a high-order nonlinear system. Besides, the iterative equations are derived by discretizing the state equations. In this way, the network´s operating efficiency is remarkably improved. This artificial neural network is designed for fitting and predicting dynamic data, and has successfully worked in simulation part of this paper.