Bayesian Adaptive Extended Kalman-Based Orbit Determination for Optical Observation Satellites.

Area of Science:

• Aerospace Engineering
• Astrodynamics
• Space Situational Awareness

Background:

• Increasing satellite numbers and space debris in Low-Earth Orbit (LEO) necessitate precise orbit determination.
• Ground-based optical observations face limitations from data constraints and noise, challenging traditional Extended Kalman Filter (EKF) accuracy.

Purpose of the Study:

• To develop a more accurate and robust method for high-precision orbit determination using optical observations.
• To overcome the limitations of the Extended Kalman Filter (EKF) in complex orbital environments.

Main Methods:

• Proposed a Bayesian Adaptive Extended Kalman Filter (BAEKF).
• BAEKF optimizes track determination via dynamic noise covariance adjustment and Bayesian a posteriori probability correction.
• Evaluated performance against traditional EKF, Unscented Kalman Filter (UKF), Radial Basis Function Neural Network (RBFNN), and Gaussian Process Regression (GPR).

Main Results:

• BAEKF demonstrated a 34.7% average reduction in root mean square error (RMSE) compared to the traditional EKF.
• Significant improvements in accuracy and stability were observed for BAEKF in nonlinear systems.
• UKF, RBFNN, and GPR also showed performance enhancements.

Conclusions:

• The proposed BAEKF offers a reliable solution for high-precision orbit determination from optical observations.
• BAEKF effectively addresses the accuracy and stability issues inherent in traditional EKF for complex space environments.
• This advancement contributes to safer spacecraft operations and enhanced space situational awareness.