"Comparative evaluation of three machine learning algorithms on improvi" by Hao Peng and Xiaoli Bai

Article Title

Comparative evaluation of three machine learning algorithms on improving orbit prediction accuracy

Authors

Hao Peng, Department of Mechanical and Aerospace Engineering, Rutgers, The State University of New Jersey, Piscataway, NJ 08854,USA
Xiaoli Bai, Department of Mechanical and Aerospace Engineering, Rutgers, The State University of New Jersey, Piscataway, NJ 08854,USA

Keywords

resident space objects (RSOs), orbit prediction, machine learning (ML), support vector regression, artificial neural network (ANN), Gaussian processes (GPs)

Abstract

In this paper, the recently developed machine learning (ML) approach to improve orbitprediction accuracy is systematically investigated using three ML algorithms, includingsupport vector machine (SVM), artificial neural network (ANN), and Gaussian processes(GPs). In a simulation environment consisting of orbit propagation, measurement, estimation,and prediction processes, totally 12 resident space objects (RSOs) in solar-synchronousorbit (SSO), low Earth orbit (LEO), and medium Earth orbit (MEO) are simulated tocompare the performance of three ML algorithms. The results in this paper show that ANNusually has the best approximation capability but is easiest to overfit data; SVM is the leastlikely to overfit but the performance usually cannot surpass ANN and GPs. Additionally,the ML approach with all the three algorithms is observed to be robust with respect to themeasurement noise.