Mining Two-Line Element Data to Detect Orbital Maneuver for Satellite

Abstract

Data clustering analysis is proposed to detect the orbital maneuvers of satellites at different scales. In this study, the unsupervised classification methods of K-means, hierarchical, and fuzzy C-means clustering are used to handle the two-line element (TLE) historical data. The K-means-based contour map method is applied to the characteristic variable selection and cluster number determination. The TLE data of large-, medium-, and small-scale orbital maneuvers are clustered by the aforementioned three methods. Through a series of numerical experiments, it is found that for different scales of orbital maneuvers, the clustering methods have different performances and that they can essentially fulfill the functional requirements of orbital detection. By data mining, the orbital maneuvers of the remote sensing satellites 'YAOGAN-9', 'TIANHUI-1', and 'Envisat' can be easily detected, which will provide useful information for further orbital supervision and prediction.