Application of Gauss-Hermite filter in SINS alignment

In this paper investigate the alignment problem of a stationary based strapdown inertial navigation system. In order to improve the aligning accuracy and shorten the aligning time, a Gauss-Hermite filter (GHF) is adopted in the alignment model based on large azimuth misalignment angles. The nonlinear Gauss integration of multi-variables to the mean and covariance computation in the GHF is addressed. Since large azimuth misalignment angles will introduce the nonlinearity in the alignment error equations, this paper employs linear state transformation approach to obtain the analytic solution of the linear state vector in the underlying equations. The integration of multi-variables is thus converted to the integration of a single-variable. Hence the so called "dimension problem" in the application of GHF to alignment is solved without loss of accuracy. The proposed method is applied to a SINS, and it shows that the aligning accuracy of path angle is improved by 16% and the aligning time is reduced by 75% compared with extended Kalman filter (EKF) and unscented Kalman filter (UKF).