Modeling and Control of Magnetic Bearing System Using Neural Networks

The principle and its approximative linear modeling of magnetic bearing system are given. The linear controller based on linear control theory and simplified modeling is developed, which results in the performance degradation and limitation of magnetic bearing system. The method of modeling and control of magnetic bearing system using nonlinear recurrent neural networks is studied,and the learning algorithm combined with the practical application of the magnetic bearing system is also developed. The proposed method restrains the effects of nonlinearity and uncertainty on the performance of magnetic bearing system,and has the ability of robustness,which improve the system performance greatly.