Enhancing working performance of active magnetic bearings using improved fuzzy control and Kalman-LMS filter

Abstract

This paper develops a novel control approach to enhance working performance of eight-pole radial active magnetic bearings (AMB). In the proposed method, the improved fuzzy proportional-integral-derivative (PID) control based on variable universes of proportional scaling is designed firstly to obtain better identification performance and flexibility of AMB control. Then, a Kalman filter is implemented to estimate the rotor displacement to reduce the noise disturbance and undesirable parameter adjustments caused by fuzzy control scheme. Finally, the least mean square (LMS) filter is connected with fuzzy control to compensate the unknown mass unbalance which might induce serious vibrations of AMB facilities. The simulation results show that the improved fuzzy PID control plays better performance than PID control in overshoot control, and the transient time of improved fuzzy PID is much shorter compared to conventional fuzzy PID. Meanwhile, the impacts of noise disturbance can be significantly limited by connecting with Kalman filter, and the maximum steady-state error can be decreased to about 85%. It is also shown that the LMS filter algorithm can effectively compensate the unknown mass unbalance in relatively short time without affecting the stability of AMB system.