Downdating a time-varying square root information filter
Abstract
A new method to efficiently downdate an estimate and covariance generated by a discrete time Square Root Information Filter (SRIF) is presented. The method combines the QR factor downdating algorithm of Gill and the decentralized SRIF algorithm of Bierman. Efficient removal of either measurements or a priori information is possible without loss of numerical integrity. Moreover, the method includes features for detecting potential numerical degradation. Performance on a 300 parameter system with 5800 data points shows that the method can be used in real time and hence is a promising tool for interactive data analysis. Additionally, updating a time-varying SRIF filter with either additional measurements or a priori information proceeds analogously.
- Publication:
-
Flight Mechanics/Estimation Theory Symposium, 1990
- Pub Date:
- December 1990
- Bibcode:
- 1990fmet.symp..291M
- Keywords:
-
- Algorithms;
- Computer Techniques;
- Covariance;
- Data Processing;
- Kalman Filters;
- Real Time Operation;
- Variations;
- Degradation;
- Detection;
- Losses;
- Astrodynamics