Algorithm of space track segment association under distributed star sensor

In view of the characteristic that the space targets can be observed intermittently by distributed star sensors, the observation information association of the space target in different time periods under star sensors is taken as the premise of accurate space object calibration based on the star sensors. The space target motion characteristic is considered.Combined with the characteristics of the space traget motion, on the basis of the previous two threshold fuzzy correlation, the track segment association algorithm of the space target under the distributed star sensor is proposed by adding the normal vector constraint of orbit plane,filtering the candidate association objects and simplifying the cost of association operation. Through the simulation,the divergence of the track extrapolation error with time is analyzed under six sets of noise levels. And the reference value of the adjustment coefficient in the fuzzy membership function is given, which makes the correlation similarity of different targets more significant under long interval. The simulation results show that the correlation accuracy of the proposed algorithm is higher than that of the nearest neighbor and the traditional fuzzy association under certain noise.When the standard deviation of the initial orbital error is 6km at each axis position and 6m/s at each axis velocity,the adjacent targets with a minimum phase difference of 0.5° can be distinguished. Correlation accuracy of interval 2h is 98%, and the correlation accuracy is 90% of 7h interval.

Conclusion

GEO orbits are distributed with a large number of spatial targets, and compared to ordinary targets, the motion of spatial targets is more regular. As spatial targets run on a specific orbital plane, the normal vector threshold constraint of the orbital plane can be used to reduce the total set of tracks to be associated to a set of tracks in a similar orbital plane, reducing the cost of target association operations. This is an improvement of the existing fuzzy association method based on the background of spatial target association in this article.
In the simulation, considering the existence of initial errors, the state estimation of spatial targets adopts forward extrapolation of old tracks and inverse extrapolation of new tracks. At the same time, correlation sample points of one hour in the middle interval are taken to reduce the impact of errors on the correlation results. In order to make the algorithm satisfy the association in a long time interval, the reference values of the adjustment parameters in the association membership function are fitted through simulation experiments, which solves the problem that the association similarity of different spatial targets in a long interval is too close.
Comparing the correlation accuracy of NN correlation, conventional fuzzy correlation, and the algorithm introduced in this paper with the orbit normal vector, the algorithm has a higher correlation accuracy under the same noise level. When the standard deviation of the initial orbit determination error at each axis position was 6km, and the standard deviation of each axis speed was 6m/s, adjacent spatial targets with a minimum phase difference of 0.5 ° could be distinguished. Within 2 hours (conventional task interval time), the correlation accuracy reached 98%, and when the correlation interval time was 7 hours, the accuracy reached 90%.