Neural Network Based Modified State Observer for Orbit Uncertainty Estimation
A novel technique for estimating uncertainties caused by gravitational perturbations is presented. The approach, called the modified state observer, allows for the estimation of uncertainties in nonlinear dynamics and, in addition, providing estimates of the system states. The observer structure contains neural networks whose outputs are the uncertainties in the system. A useful and important application of this observer is the problem of determining uncertain gravitational perturbations that a satellite may experience when orbiting a body. With future space missions involving other bodies, such as asteroids that produce gravitational perturbations which are highly uncertain and are subjected to unknown physical influences, the modified state observer can be used not only to estimate the states of the satellites, but it can also be used to estimate the uncertainties that could be analyzed further with understanding the physical phenomena. To demonstrate the utility of the modified state observer for this class of problem the technique is applied for two cases: estimating the uncertainty caused by the J2 perturbation for an Earth orbiter and estimating the uncertainty in an asteroid's gravitational field. Simulations are presented, which indicate that the observer can accurately estimate both the periodic nature of these perturbations, as well as the magnitudes.
N. Harl et al., "Neural Network Based Modified State Observer for Orbit Uncertainty Estimation," Journal of Guidance, Control, and Dynamics, vol. 36, no. 4, pp. 1194-1209, American Institute of Aeronautics and Astronautics (AIAA), Jul 2013.
The definitive version is available at https://doi.org/10.2514/1.55711
Mechanical and Aerospace Engineering
International Standard Serial Number (ISSN)
Article - Journal
© 2013 American Institute of Aeronautics and Astronautics (AIAA), All rights reserved.