GMM-Based Adaptive Extended Kalman Filter Design for Satellite Attitude Estimation under Thruster-Induced...

Area of Science:

• Aerospace Engineering
• Control Systems
• Signal Processing

Background:

• Star trackers are crucial for satellite attitude determination, but maneuvering causes defocused star images and degrades centroid measurement accuracy.
• Geostationary satellite station-keeping maneuvers, involving thruster misalignment and impulsive torque, induce non-Gaussian noise in star images, complicating attitude estimation.
• Traditional Extended Kalman Filters (EKF) struggle with non-Gaussian noise, leading to inaccurate star vector outcomes.

Purpose of the Study:

• To develop and evaluate an improved filtering method for accurate satellite attitude estimation, specifically addressing non-Gaussian measurement noise during maneuvers.
• To enhance the prediction of measurement vectors by incorporating a Gaussian Mixture Model (GMM) within an adaptive Extended Kalman Filter framework.

Main Methods:

• Implementation of an adaptive Extended Kalman Filter (EKF) to predict measurement vectors using predicted states.
• Integration of a Gaussian Mixture Model (GMM) to assign weighting parameters to Gaussian densities, improving measurement vector prediction.
• Simulation-based performance comparison between the proposed GMM-EKF and the standard EKF for attitude estimation.

Main Results:

• The GMM-EKF demonstrated superior performance compared to the standard EKF in satellite attitude estimation.
• A significant 30% improvement in attitude estimation performance was achieved using the GMM-EKF.
• The GMM-EKF effectively handles non-Gaussian noise inherent in star images during satellite maneuvers.

Conclusions:

• The GMM-EKF offers a more robust and accurate approach for attitude estimation in geostationary satellites undergoing station-keeping maneuvers.
• This method provides a viable solution for mitigating the adverse effects of non-Gaussian measurement noise on attitude determination.
• The enhanced prediction capabilities of the GMM-EKF make it an attractive alternative for improving satellite navigation and control.