Adaptive model Predictive control for a magnetic suspension system under initial position dispersions and voltage disturbances

Abstract

A nonlinear magnetic suspension system consisting of a mechanical motion and electromagnetic circuit is considered in this paper. An online algorithm that is used to stabilize the suspended mass targeting the desired ball position, ball velocity and coil current is presented. A steady-state condition, in which the ball position, the ball velocity, and the coil current were assumed constant, was used as a reference trajectory for the closed-loop simulations. An adaptive model predictive control method was employed to control the coil voltage while the discrete plant model and operating conditions were changed at each time step. The effectiveness of the proposed control law was validated in the presence of disturbances in initial ball position, steady-state ball position, suspended mass, and voltage using the Monte-Carlo simulation method. The sinusoidal, step, and impulse voltage disturbances were applied consequently to the system while imposing random dispersions of the initial ball position. Numerical results with five hundred trials illustrated that the designed algorithm can stabilize the system and track the desired reference without exceeding the state and input constraints despite the wide range of dispersions in initial ball position, steady-state ball position, suspended mass, and input voltage disturbances.