Accession Number:

AD1096796

Title:

Advanced Orbit Prediction for Resident Space Objects through Physics-based Learning

Descriptive Note:

Technical Report,15 Apr 2016,14 Apr 2019

Corporate Author:

Rutgers, The State University of New Jersey New Brunswick United States

Personal Author(s):

• Bai,Xiaoli

Report Date:

2019-07-11

Pagination or Media Count:

7.0

Abstract:

The goal of this research is to develop a novel methodology to predict trajectories of resident space objects RSOs with orders-of-magnitudeshigher accuracy than the current methods. We propose to enhance physics-based orbit prediction with a learning-based system identification well suited for the challenging, unstable, and inactive RSOs that are out of control and have uncertain origins. We have developed a simulation-based space catalog environment to validate the proposed orbit prediction method. For the first time, our simulation results demonstrated three types of generalization capability for the proposed approach. We have also validated the developed ML methodology using publicly available data.

Descriptors:

• resident space objects
• dynamics
• trajectories
• accuracy
• machine learning
• space situational awareness
• algorithms
• Gaussian Processes
• neural nets

Subject Categories:

• Astronomy
• Statistics and Probability
• Cybernetics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE