Purpose <p>To address the multi-source horizontal vibration issues in high-speed elevator car systems (including random disturbances induced by rail surface wear and piston wind in the hoistway, as well as transient impacts caused by uneven rail joints), this paper proposes a filter-based adaptive control strategy for electro-hydraulic active guide shoes. </p> Methods <p>A nonlinear eight-degree-of-freedom dynamic model of the high-speed elevator car system and an electro-hydraulic active guide shoe dynamic model are established. An adaptive backstepping controller is designed to address the nonlinear characteristics of the car system. To resolve the "complexity explosion" problem caused by high-order derivative terms in the backstepping control method for the elevator car system, the desired output force of the electro-hydraulic system is filtered, and an adaptive control method considering the dynamic performance of electro-hydraulic active guide shoes is proposed. A composite Lyapunov function is constructed within the backstepping framework to prove the global stability of the closed-loop system rigorously. </p> Results <p>Under both random and pulse rail excitations, the proposed controller (considering guide shoe dynamics) is compared with standard adaptive backstepping control (without considering guide shoe dynamics) and passive control. Results demonstrate that the proposed method reduces the peak horizontal vibration acceleration by over 68% and the root mean square (RMS) value by over 75% compared to standard adaptive backstepping control, effectively suppressing horizontal vibrations and validating the efficacy of the proposed approach. </p> Conclusion <p>Comprehensive evaluations (including robustness against sensor noise, energy consumption analysis, and parameter sensitivity of the filter time constant) were systematically conducted to fully verify the engineering feasibility and practical performance of the proposed control strategy.</p>