Date of Award
Master of Science in Computer Science
Department of Electrical and Computer Engineering
Laurence D. Merkle, PhD
Ground-based non-resolved optical observations of resident space objects (RSOs) in geosynchronous orbit (GEO) represent the majority of the space surveillance network’s (SSN’s) deep-space tracking. Reliable and accurate tracking necessitates temporal separation of the observations. This requires that subsequent observations be associated with prior observations of a given RSO before they can be used to create or refine that RSO’s ephemeris. The use of astrometric data (e.g. topocentric angular position) alone for this association task is complicated by RSO maneuvers between observations, and by RSOs operating in close proximity. Accurately associating an observation with an RSO thus motivates the use of photometric light curves in that association process. Contemporary machine learning, specifically deep neural networks (DNNs), offers mechanisms to perform that association autonomously by first learning patterns between observations and their parent RSOs. This research assesses the extent to which a trained DNN can autonomously associate previously unseen observations with the RSO they represent. DNN performance is assessed by recording the percentage of observations from a held-out testing set associated to the correct RSO. Model performance is evaluated in three object-observation association scenarios: 1) no maneuvering, 2) objects maneuver in progressively closer proximity, and 3) objects permute stations with one another. In each simulation, photometric observations are generated as if observed by the Maui GEODSS site. This research contributes a foundational proof of concept for object classification via photometric signature, and contributes to the development of autonomous SSN telescopes, systems to distinguish closely spaced RSOs, and systems to autonomously and rapidly update an RSO’s ephemeris after it maneuvers.
DTIC Accession Number
McQuaid, Ian W., "Autonomous Association of GEO RSO Observations Using Deep Neural Networks" (2019). Theses and Dissertations. 2270.