AbstractReaction wheels, as one of the most commonly used actuators in satellite attitude control systems, are prone to malfunction which could lead to catastrophic failures. Such malfunctions can be detected and addressed in time if proper analytical redundancy algorithms such as parameter estimation and control reconfiguration are employed. Major challenges in parameter estimation include speed and accuracy of the employed algorithm. This paper presents a new approach for improving parameter estimation with adaptive unscented Kalman filter. The enhancement in tracking speed of unscented Kalman filter is achieved by systematically adapting the covariance matrix to the faulty estimates using innovation and residual sequences combined with an adaptive fault annunciation scheme. The proposed approach provides the filter with the advantage of tracking sudden changes in the system non-measurable parameters accurately. Results showed successful detection of reaction wheel malfunctions without requiring a priori knowledge about system performance in the presence of abrupt, transient, intermittent, and incipient faults. Furthermore, the proposed approach resulted in superior filter performance with less mean squared errors for residuals compared to generic and adaptive unscented Kalman filters, and thus, it can be a promising method for the development of fail-safe satellites.
HighlightsFaults in Reaction wheels are examined.A high fidelity model of a Reaction Wheel is considered.An Enhanced Adaptive Unscented Kalman Filter is presented.Adaptation is based on innovation and residual sequences and statistical distribution theories.Results show superior performance compared to unscented Kalman filters.
Fault estimation of satellite reaction wheels using covariance based adaptive unscented Kalman filter
Acta Astronautica ; 134 ; 159-169
2017-02-03
11 pages
Article (Journal)
Electronic Resource
English
Residual Based Adaptive Unscented Kalman Filter for Satellite Attitude Estimation
British Library Conference Proceedings | 2012
|Aerodynamic parameter estimation using adaptive unscented Kalman filter
Online Contents | 2013
|