Taxonomies are useful for providing structure when categorizing large numbers of near-Earth space objects by the types of maneuvers they perform. In particular, lateral thrusting, longitudinal thrusting, and drifting may be directly linked to detectable changes in the orbital elements that affect object location and orientation. The purpose of this work is to develop a fully-automated taxonomy of the geosynchronous objects based on dynamical principles. Groups of objects are found using clustering methods, and two clustering methods are compared for constructing the taxonomy. The first is an adaptive k-means algorithm that does not require a priori information. It is compared to an agglomerative clustering algorithm that utilizes limits on cluster sizes to form distinct clusters. The effectiveness of the automated taxonomy is determined by comparison with the European Space Agency’s DISCOS database and clusters from the Geosynchronous yearly report.

分类法对于按结构执行的操作类型对大量近地空间物体进行分类时提供结构很有用。特别地，横向推力，纵向推力和漂移可以直接与影响物体位置和方向的轨道元件中可检测到的变化联系在一起。这项工作的目的是基于动力学原理开发地球同步对象的全自动分类法。使用聚类方法找到对象组，并比较两种聚类方法来构建分类法。第一种是不需要先验的自适应k均值算法信息。将其与利用集群大小限制形成不同集群的聚集集群算法进行比较。通过与欧洲航天局的DISCOS数据库和地球同步年度报告中的集群进行比较，可以确定自动分类法的有效性。