On control structure scheme for nonlinear system based on ANN feedforward/inverse models and simulation

Identify and modelling based on system´s inverse model was an important method to nonlinear system. In order to model nonlinear system with complex, unknown and the indefinite characteristics, the nonlinear system need reversible, reasonable modelling structure scheme and inverse model with good performance. Basing on Artificial Neural Network (ANN) nonlinear mapping and so on good performance, the modelling control structure scheme based on ANN´s inverse model had been designed. Used Exponential Forgetting and Resetting Algorithm (EFRA), the ANN´s inverse model had been trained on-line, and the ANN´s inverse model with satisfied identification request had been gotten. After massive simulation researches have been done, it is proved the designed control structure scheme is reasonable, the EFRA is effective to raise effect of identification and modelling and control performance to the nonlinear system.