Self-Sensing Three-Pole Magnetic Bearing Using a Kalman Filter

Abstract

Self-sensing active magnetic bearing is designed by a new approach. The approach aims to solve the observer bias problem by introducing a Kalman filter. In order to verify the validity of the approach, a Kalman filter is designed for a three-pole homopolar magnetic bearing and used to estimate the radial displacement and velocity of the rotor. The electric current is driven by a linear power-amplifier circuit to flow through the coils, and the coil terminal voltage is passed through an analog first-order low-pass filter with a cut-off frequency of 100 Hz. The designed Kalman filter uses the filtered coil-voltage and the controller inputs as an input for calculating the estimate. The unobservable bias is estimated as an unknown state, and the Kalman-filter estimates are numerically simulated by the measured input/output data. The results show that the bias is successfully estimated to overlap the estimated displacement to its measurement