Maglev train levitation systems are the core components that determine the safety, smoothness, and operational efficiency of maglev trains. However, their inherent open-loop instability, strong electromagnetic nonlinearity, and sensitivity to external disturbances severely hinder the achievement of high-precision levitation control. To address these critical defects, this paper proposes a data-driven sliding mode observer-enhanced model-free adaptive control (DSO-Enhanced MFAC) algorithm. Its key contribution lies in the integration of a data-driven sliding mode observer (DSO), which not only accelerates adaptive adjustment speed and enhances anti-noise performance but also eliminates the reliance on accurate mathematical models of maglev systems while significantly boosting overall control robustness. Targeted simulations, designed to mimic real-world operating conditions of maglev ball systems, fully verify the effectiveness of the proposed algorithm in ensuring stable and precise levitation.