Loading...

Entropy-based adaptive attitude estimation

Kiani, M ; Sharif University of Technology | 2018

558 Viewed
  1. Type of Document: Article
  2. DOI: 10.1016/j.actaastro.2017.12.044
  3. Publisher: Elsevier Ltd , 2018
  4. Abstract:
  5. Gaussian approximation filters have increasingly been developed to enhance the accuracy of attitude estimation in space missions. The effective employment of these algorithms demands accurate knowledge of system dynamics and measurement models, as well as their noise characteristics, which are usually unavailable or unreliable. An innovation-based adaptive filtering approach has been adopted as a solution to this problem; however, it exhibits two major challenges, namely appropriate window size selection and guaranteed assurance of positive definiteness for the estimated noise covariance matrices. The current work presents two novel techniques based on relative entropy and confidence level concepts in order to address the abovementioned drawbacks. The proposed adaptation techniques are applied to two nonlinear state estimation algorithms of the extended Kalman filter and cubature Kalman filter for attitude estimation of a low earth orbit satellite equipped with three-axis magnetometers and Sun sensors. The effectiveness of the proposed adaptation scheme is demonstrated by means of comprehensive sensitivity analysis on the system and environmental parameters by using extensive independent Monte Carlo simulations. © 2018 IAA
  6. Keywords:
  7. Confidence level ; Cubature Kalman filter ; Innovation-based adaptive estimation ; Adaptive filtering ; Adaptive filters ; Covariance matrix ; Entropy ; Intelligent systems ; Ion beams ; Monte Carlo methods ; Orbits ; Sensitivity analysis ; Adaptive estimation ; Attitude estimation ; Confidence levels ; Cubature kalman filters ; Relative entropy ; Kalman filters
  8. Source: Acta Astronautica ; Volume 144 , 2018 , Pages 271-282 ; 00945765 (ISSN)
  9. URL: https://www.sciencedirect.com/science/article/pii/S0094576517309761