Real-Time Application of Robust Offset-Free MPC in Maglev Planar Machine

Abstract

This paper improves the transient response of a magnetically levitated (maglev) planar machine through the model predictive control (MPC) benefiting from explicitly incorporating the tightened constraints of the input vector. The maglev system is characterized as a second-order linear model via the inverse model method to facilitate the fast online receding horizon optimization. The estimated lumped disturbance compensates for the deviations of the prediction caused by the unmodeled errors, so the obtained control signals eventually eliminate the steady-state errors. The robust stability of the motion control system is guaranteed by a reasonable horizon length without requiring the state constraints, which reduces the computational burden for the real-time implementation. Experimental results demonstrate that the proposed controller outperforms the anti-windup PID controller, switching linear quadratic regulator, baseline MPC, and nonlinear MPC in terms of the trajectory tracking.