The Effect of Kalman Filtering Initial Conditions on SINS Alignment

Jia-he XIA,Yong-yuan QIN,Rui LONG
DOI: https://doi.org/10.3873/j.issn.1000-1328.2010.08.007
2010-01-01
Abstract:It is difficult to correctly specify the initial condition of Kalman filter in strapdown inertial navigation system (SINS) alignment. The effect of initial conditions on SINS alignment accuracy is analyzed. The Kalman filter estimation error due to incorrect initial condition is analyzed. The relation of the real mean square error matrix (MSEM) and the MSEM computed by the filter are deduced. The Kalman filter and the least square estimation are compared for regression coefficients estimation, and a sufficient condition that Kalman filter is better than least square estimation under incorrect initial condition specification is given. Simulation for SINS static alignment shows that a suitable initial MSEM is favorable to Kalman filtering state estimation, especially in the filter's start-up stage. It is reasonable to select a small upper bound of the real MSEM as the initial filter MSEM.
What problem does this paper attempt to address?