基于线性DBSCAN聚类的空间群目标实时提取方法

Real-time extraction method of space group target based on linear DBSCAN clustering

为有效应对低轨巨型星座大规模部署,入轨早期卫星集群分布,难以对空间群目标实现有效编目管理问题,提出了一种基于线性DBSCAN聚类的空间群目标实时提取方法,并给出基于观测数据时序向量实时聚类的空间群目标判别方法.采用星链星座的TLE模拟空间群目标数据开展算法验证,实验结果表明,线性DBSCAN聚类算法对空间集群目标分类识别成功率和准确率达99%以上,识别效率较DBSCAN算法提高了约1/3;线性DBSCAN聚类算法鲁棒性强、识别准确,计算效率高,对空间群目标分辨具有较高价值.

In order to effectively deal with the large-scale deployment of low-orbit giant constellation distributed in clusters in the early stage of orbit,which makes it difficult to achieve effective cataloging management of space group targets,this paper proposes a real-time extraction method of space group targets based on linear DBCAN clustering,and also presents a space group target discrimination method based on temporal vector real-time clustering of observed data.The TLE simulated space group target data of star-link constellation is used to carry out the algorithm verification,and the experimental results show that the success rate and accuracy of linear DBSCAN clustering algorithm for space cluster target classification are more than 99%,and that the linear DBSCAN clustering algorithm has strong robustness,accurate recognition and high computational efficiency,which is of high value for space group target recognition.