The Identification Model of Magnetic Bearing Supporting System

The neural networks identification model is developed on the basis of the force analysis of magnetic bearing spindle, which reflects the nonlinear delay character between inputs and outputs system. This network is able to converge quickly in 5 training steps. The mean square error value reduces to 3.495e-006 in 50 steps. Inspection shows that the neural networks DTNN identification model can fit the I/O character of the magnetic bearing supporting system within a permitted error range. This paper proposes a new approach for magnetic bearing system modeling.