Space object characterization and classification using multiple model approach

Abstract

This thesis addresses the research on characterization and classification of a space object at low Earth orbit using a multiple model approach. Space situational awareness has become a global interest as many industries started diving into the space market for commercialization of the resident space. The states of a space object can be estimated based on the orbital mechanics, but the inference of the object's characteristics is difficult while it is necessary when there is no a priori information about the detected object that its re-entry trajectory needs to be estimated. Thus, this thesis introduces an algorithm that estimates the ballistic coefficient of an object and classifies into a catalog depending on its shape with the multiple-model adaptive estimation that uses a parallel set of Kalman filters with multiple models. The simulation is done at the altitude of 300 km where the re-entry is expected within approximately two months, and the increase in accuracy of ballistic coefficient estimate through the moving-bank MMAE is shown. Also, the feasibility of supplementing the lack of data using techniques that speed up the convergence in the algorithm is presented.