The determination of unobservable states is important in consideration of system performance during initial alignment.
The aim of observability analysis is concluded as follows: 1) finding the observable states or linear combinations of these
states; 2) finding those states whose measurements turn the system into a completely observable; 3) separating the
system into observable and unobservable subsystems. In this paper, the system equation and measurement equation of
SINS for Kalman filtering are given. The Observability of initial alignment process of SINS is analyzed by means of
singular value decomposition method. Degree of observability for every state can be computed by preceding method, the
three unobservable states of INS are obtained; therefore optimal observable subspace is determined by structure
decomposition method. For proving the correctness and effectiveness of this proposed method, a Kalman filter is
designed. The Kalman filtering results are obtained. By comparing these results with results of observability analysis, the
uniform conclusion is obtained. Before designing a Kalman filter, the observability of every systematic state can be
known by use of the singular value decomposition method of the observable matrix, i.e. degree of observability for every
state can be computed. Therefore observable vector and optimal observable subspace may be determined. The
correctness and effectiveness of this proposed method was proven by analyzing results of Kalman filtering. This method
can be used for direct design of a Kalman filter.
|