Design-Oriented Benchmarking of Sliding Mode Controllers for a Magnetic Levitation System Under Actuator and Sensor Constraints
This study examined the control of a single-axis vertical magnetic levitation (MagLev) system under realistic implementation constraints. The system was considered due to its nonlinear dynamics and inherent open-loop instability, which make it suitable for evaluating practical control limitations. Actuator voltage and current saturation, parameter uncertainty, external disturbances, and measurement noise were explicitly incorporated into the analysis. For comparative evaluation, three control strategies were implemented: local linear state feedback control based on an LQR design, fixed-gain sliding mode control, and adaptive sliding mode control. Rather than relying on a single initial condition, controller performance was assessed over ranges of initial positions and velocities using a capture region framework. The effects of parametric uncertainty were systematically investigated through Monte Carlo–based numerical simulations. The results indicated that the linear controller exhibited zero successful captures over the evaluated initial condition grid, while sliding mode controllers achieved restricted but nonzero capture regions under nominal and disturbance conditions. The adaptive mechanism influenced transient behavior but did not substantially expand the feasible operating region due to actuator constraints. Measurement noise was found to degrade controller effectiveness in all cases. These findings emphasize the importance of jointly considering physical constraints and stability properties when evaluating control strategies for magnetic levitation systems.