Simulation Study of Identification and Modelling for State Equation Model Based on Neural Network

The principle of system´s identification and modelling based on the artificial neural network (NN) were analysed. In order to avoid complex description and improve simple of algorithmization for the models of dynamic systems, the models of state equation had been adopted, and the identification models of state equation for dynamic systems based on the NN were given out. For comparing to identification effect and generalization ability of different structure of the NN´s models, simulation researches had been done to three kinds of different network structure schemes of state equation models for the dynamic systems. Simulation results show, the identification models of state equation for dynamic systems based on the NN are effective, and the simply reasonable structure of the NN models can be raised the generalization ability of the NN models.