Nonlinear Nonparametric Identification of Magnetic Bearing System Based on Support Vector Regression

The mathematical model of components in an active magnetic bearing system plays important roles in system designing and adjusting. The linearized models are widely applied; however, sometimes these models are not precise enough. In this paper, a nonlinear modeling method based on Support Vector Regression (SVR) is proposed. The proposed model treats the rotor acceleration as a nonlinear function of the rotor displacement, rotor velocity and the coil currents; the model parameters are determined by a training procedure based on observed samples, no mechanism, computational or empirical models are needed in the modeling procedure. The proposed method is validated by a series of experiments.