Support vector machine-based high-precision orbit prediction for LEO satellites

To improve the orbit prediction accuracy of Low Earth Orbit satellites (LEO), this study proposes an orbit prediction method based on Support Vector Machine (SVM). Taking the XJS-G satellite as the research object, this paper explores the construction of a specific data set using three key features: position, velocity, and orbital elements. The model is trained by adopting training sets of different scales. The research results show that an increase in the scale of the data set is conducive to enhancing the prediction effect, but there is a threshold effect in terms of performance improvement. This paper successfully constructs a dataset with good prediction results, clarifies the influence of the dataset size on the prediction performance of the model, and provides new ideas for subsequent related research and aerospace application fields.