Robust implementation of Kalman filter for INS correction Yuri Kim, Anton de Ruiter,
James Lee, Alfred Ng Canadian Space Agency This paper discusses the correction
problems [1-8] and the two most general methods of implementation of the Kalman Filter (KF) for Inertial Navigation System (INS)
correction with an external Navigation Aids; correction errors outside of INS
by analytical compensation and physical correction (including alignment) by
closed loop negative feedback control. It is shown that the scheme of closed
loop control is more preferable from an integrated system robustness point of
view as well as from the point of view of computations economy. Two
methods of implementing the Kalman Filter for INS
error correction have been analyzed from the point of view of system
robustness. Both schemes result in identical mathematical descriptions for the
state estimation errors. While the feed forward compensation scheme has the
benefit that both the INS and the external navigation aid may be implemented
with only the need for the output from each (no input to each is required), the
disadvantage is that such an implementation is not robust to modeling errors.
The feedback scheme on the other hand requires that the user be able to provide
an input to the INS. This scheme has the advantage that it is robust to
modeling errors. Furthermore, it turns out that the feedback scheme is more
computationally economical, requiring only the implementation of the filter
gain matrix
The
details of the implementation of both schemes have been demonstrated by a
simple example of INS vertical channel correction using an altimeter. |
© 1995-2008 Kazan State University
