In this paper, a position regulation control strategy is developed for a magnetic levitation system operating in the presence of a bounded, periodic disturbance. Specifically, the proposed controller utilizes a saturated control force input in conjunction with a learning-based disturbance estimator to asymptotically regulate the target mass to a desired set point position despite the actuator's unidirectional limitation of exerting only an attractive force on the target mass (i.e., the control input can only generate an attractive force while the earth's gravitational field is utilized to produce the repulsive action). Differing from the previous approaches, the learning-based estimator only requires that the disturbance be bounded and that the period of the disturbance be known (i.e., the structure of the disturbance is not required to be known). Experimental results are included to illustrate the performance of the control strategy