Abstract: Aiming at the problem of degraded speed tracking performance of the hydraulic roofbolter rotary system caused by time-varying parameters, external load disturbances, and repetitive operation uncertainties during the drilling process, an observer-based model-free adaptive iterative learning control method is proposed by leveraging the periodic repetitive characteristics of drilling operations. First, a dynamic model of the rotary system is established based on the flow-pressure relationship of the electro-hydraulic proportional control system, and the system is described as a nonlinear iterative form with non-repetitive uncertainties by incorporating the repetitive operation characteristics of the drilling process. Second, an iterative observer is designed for the estimation and compensation of lumped uncertainties, which updates the system output and uncertainty estimates using historical iterative observation errors. On this basis, the observation information is introduced into the parameter update law and the iterative learning control law to reduce the influence of lumped uncertainties on the data-driven learning process and improve the disturbance rejection performance of the system under variable working conditions. By analyzing the recursive relationship between the tracking error and the observation error, the bounded convergence of the speed tracking error along the iteration axis is proved. Finally, simulation verification is carried out on the MATLAB-AMESim co-simulation platform. The results show that the proposed method can improve the speed tracking accuracy and iterative convergence performance, and exhibits good robustness against load variations and lumped uncertainties.