A Fast Association Method for Angular Trajectories Based on Siamese Long Short Term Memory Network

With the rapid growth in the number of space objects, the utilization of space-based optical systems for space object positioning holds significant importance for space security. As a critical component in multi-satellite collaborative positioning, angular trajectory association faces challenges with traditional geometry or kinematics-based methods, including combinatorial explosion, noise sensitivity, and poor real-time performance. This paper proposes a deep learning-based fast angular trajectory association method. By constructing multi-satellite multi-target simulation scenarios, simulated angular trajectory data are generated, and a Siamese long short-term memory network is designed for angular trajectory association. Experimental results demonstrate that, compared with traditional methods, the proposed method improves accuracy by about 3% and reduces inference time by more than 9 times. Therefore, this method can significantly increase the association speed of space multi-target trajectories while maintaining high accuracy, providing a reference for research on space objects’ trajectory association.