Satellite attitude estimation using SVD-Aided EKF with simultaneous process and measurement covariance adaptation

Chingiz Hajiyev, Demet Cilden-Guler*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

13 Citations (Scopus)

Abstract

The attitude estimation of a spacecraft in low Earth orbit is considered with the design of two different adaptation rules in the extended Kalman filter (EKF) algorithm. The adaptations are designed for compensating both the measurement faults and external disturbances by updating the noise covariances of the Kalman filter. First, the measurement noise covariance (R) adaptation is introduced by using the Singular Value Decomposition (SVD) as a preprocessing step in EKF design. The estimation filters might suffer from the large erroneous initialization of the states by diverging from the actual case. The proposed algorithm on the other hand uses SVD measurements as the initial conditions for the filtering stage. This makes the filter resistant to this type of error source. Second, the process noise covariance (Q) adaptation rule is incorporated into the previous filter design. The rules are set simultaneously so that the filter has the capability to be robust against initialization errors, system noise uncertainties, and measurement malfunctions. Numerical simulations based on several scenarios are employed to investigate the robustness of the filter.

Original languageEnglish
Pages (from-to)3875-3890
Number of pages16
JournalAdvances in Space Research
Volume68
Issue number9
DOIs
Publication statusPublished - 1 Nov 2021

Bibliographical note

Publisher Copyright:
© 2021 COSPAR

Funding

D. Cilden-Guler is supported by ASELSAN and TUBITAK PhD scholarships.

FundersFunder number
TUBITAK
Aselsan

    Keywords

    • Adaptive filtering
    • Attitude estimation
    • Extended Kalman filter
    • Rate gyro
    • Singular value decomposition
    • Small satellite

    Fingerprint

    Dive into the research topics of 'Satellite attitude estimation using SVD-Aided EKF with simultaneous process and measurement covariance adaptation'. Together they form a unique fingerprint.

    Cite this