Position Estimation for Self-sensing Magnetic Bearings using Artificial Neural Network

Although signal injection, state observer and parameter estimation methods were previously studied for selfsensing AMBs, variety of nonlinear effects such as eddy current, magnetic saturation and coil flux leakage make it difficult to apply for industrial applications. In this paper, we study a position estimation for self-sensing active magnetic bearings (AMBs) using artificial neural network (ANN), especially RNN method. Mathematical model of self-sensing AMBS are introduced including PWM duty, average current, current ripple and current slope, and various nonlinear effects are investigated quantitatively. We applied ANN method to deal with the non-linear effects of self-sensing AMBs. Finally, selfsensing AMBs using ANN are simulated by MATLAB Simulink its performances are compared with previous self-sensing methods.