Nonlinear Coupled Dynamics, Fuzzy Neural-Adaptive Control, and Safety Thresholds of High-Speed Maglev Trains under Guideway Disturbances

High-speed maglev trains operating on flexible guideway bridges are sensitive to localized track steps and pier settlements, which can drive the electromagnetic system into saturation and amplify coupled train-guideway vibration. This study develops an integrated framework combining a saturation-corrected electromagnetic-force model, fuzzy neural-network adaptive backstepping control (FNN-ABSC), and coupled structural-response analysis. The force model is calibrated using COMSOL results and validated against published experimental measurements; errors remain within 6% and decrease to below 3% in the saturated range. The FNN-ABSC uses gap-error and gap-rate information to preserve levitation-gap tracking while suppressing rapid transients. Relative to Fuzzy-ABSC, it reduces settling and rise times by approximately 32.4% and 21.2%, respectively, with zero steady-state error and no overshoot. In the coupled model, identical controller instances share the network architecture and initial parameters but adapt independently online using local signals. The measured track-irregularity profile is supplemented with localized deterministic track steps and pier-settlement defects. Track steps introduce broadband transient excitation and strongly amplify local electromagnet, levitation-frame, and guideway accelerations, whereas the smoother long-wavelength settlement excites comparatively stronger low-frequency carbody bounce/pitch response. Under the measured baseline irregularity, guideway mid-span acceleration increases from 0.15 m/s 2 at 120 km/h to 0.95 m/s 2 at 600 km/h. For the studied configuration at 600 km/h and the adopted 5-15 mm absolute safety band, the conservative tolerable track-step and pier-settlement amplitudes are approximately 4 mm and 18 mm, respectively.