Maglev electromagnetic suspension (EMS) under dynamic operation is challenged by track-induced disturbances and modeling uncertainties, leading to levitation-gap oscillations and steady-state bias. A semi-bogie EMS model is developed, and Lyapunov-based uniform ultimate boundedness under bounded disturbances is established, explicitly linking controller gains to the uniform ultimate bound and convergence rate. A dual-path disturbance-compensated sliding-mode controller is proposed: a feedforward predictor pre-cancels foreseeable disturbances, whereas an observer estimates and rejects residual lumped disturbances online. The two paths are adaptively fused via online weighting to form FDA-SMC. SIMPACK–SIMULINK co-simulations show that all compared controllers satisfy the exceedance-probability criterion. FDA-SMC confines the peak gap to 12.20 mm with a maximum exceedance probability of 0.15% and achieves zero exceedance at the right-side suspension locations. Sliding-variable statistics indicate up to 29% and 25% reductions in mean magnitude and RMS versus baseline SMC, demonstrating improved robustness under multi-source, frequency-varying perturbations.
Fei et al. (Thu,) studied this question.