Abstract:

In view of the real-time tracking problem of space non-cooperative spatial targets such as constellation groups of giant stars, a space-based distributed cooperative passive detection method for multi-target tracking is proposed. Firstly, a dynamic model of satellite orbit under the conditions of Earth's non-spherical J2 perturbation and atmospheric drag perturbation is established, along with a unit line-of-sight vector measurement model onboard. Then, a multi-target tracking algorithm based on cardinalized probability hypothesis density filtering is developed, and an approximate closed-form solution to multidimensional integrals is obtained using the Gaussian mixture method to reduce computational complexity and address onboard implementation issues. Furthermore, a consistency information fusion scheme for multi-platform multi-target tracking interaction is designed, incorporating labels for target discrimination to mitigate computational matching issues arising from information exchange and fusion between different platforms, and employing consistency information filtering scheme for information fusion. Finally, the proposed method is validated through simulation experiments using 15 orbitally proximate constellation satellites in a local region of a constellation as tracking targets. Simulation results demonstrate the effectiveness of the proposed method, with a tracking performance improvement of approximately 60% compared to traditional methods, and position tracking errors within 5 km under non-singular cooperative configurations.

Key words: passive detection, multi-target tracking, consensus-based information filtering, distributed collaborative observation