Skip to Main Content
For realization effective identification and modelling and implementation control to complex and unknown nonlinear systems, firstly, analysis and research were done for structure schemes of identification and modelling and control based on inverse models. Secondly, based on characteristics of nonlinear approach and adaptive study of artificial neural network (ANN) and so on, using the methods of the generalized training and specialized training for ANN inverse model, simulation studies were done for the BP MFN (multilayer feedforward network) inverse model based on the EF(exponential forgetting) renews covariance matrix algorithm. After full iteration training, the inverse model network which the structure had been optimized was obtained, and it may use the structure scheme of identification and modelling and control for nonlinear system. Simulation results show that, using appropriate structure of the inverse model network designed, after full training, identification and modelling and control are effective for nonlinear system.