Satellite Orbit Propagator (SOP) is of prime importance in the prevention of collision and completion of the assigned task of the satellites. In the past, orbit prediction and propagation have relied on physics-based mathematical models. However, as the number of satellites and their data increases, it is crucial to explore the data-driven orbit propagation based on advanced machine learning methods. In this work, we propose a novel deep learning-based framework to forecast future satellite orbit states. The proposed framework employs a model based on Neural Controlled Differential Equations (NCDEs) to train orbit prediction models, and our approach captures features from past satellite state values at both fixed and dynamic time intervals. The experimental results on Korea Aerospace Research Institute (KARI)’s KOMPSAT-3 and 5 datasets demonstrate that the proposed framework outperforms the other eight data-driven baseline forecasting models.