A Control Theoretic Framework For Accurate Tracking Of A Satellite In An Unknown Environment
Open Access
- Author:
- Mc Shane, Richard
- Area of Honors:
- Aerospace Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Puneet Singla, Thesis Supervisor
Kenneth Steven Brentner, Thesis Honors Advisor - Keywords:
- Resident Space Objects
Estimation
Filter
Nonlinear Predictive Filter
Astromechanics
Atmospheric Drag
Low-thrust Maneuver
Minimum Model Error
Optimum Control Problem
mis-model
State Transition Matrix
covariance constraint
Orbital Mechanics
Orbit Estimation
Dynamic Mis-modeling
Observation
sensor
Space Situational Awareness
Adaptive Filter
control
Estimated States
Orbit
nonlinear control
nonlinear filter
predictive filter - Abstract:
- In Space Situational Awareness (SSA), there is a need to estimate the orbit states of non- cooperative resident space objects (RSOs) in a data sparse environment. By combining knowledge of the state dynamics and observations from a sensor, one is able to estimate the RSO orbit state. Many benchmark estimators, such as the Kalman filter, require accurate dynamic models and are prone to any modeling errors. This can be a limiting factor in SSA as RSOs motion dynamics may include unknown external forces such as atmospheric drag and/or unknown maneuvers. There is a need for methods to estimate orbit states of an RSO accurately in presence of modeling errors. Implementing these filters will be the first step in understanding, improving, and creating new filtering algorithms. In this work, a nonlinear predictive filter is defined and implemented. This filter estimates the mis-modeling error in an optimal control framework such that the measurement residuals match known sensor noise characteristics. The validation studies include an RSO in low Earth orbit under the effect of unknown atmospheric drag and RSO going through unknown low thrust maneuver. The results obtained shows the efficacy of the predictive filter in providing accurate estimates of the orbital states of an RSO while the estimates of modeling errors are shown prone to different tuning parameters.