Abstract:

In order to improve alignment precision and reduce time for the initial alignment of strapdown inertial navigation system (SINS) under the moving base, a new method that nonlinear alignment with 5th-degree cubature Kalman filter (CKF) without coarse alignment is proposed. Because of lower accuracy of 3rd-degree CKF and too much calculation work of the 5th-degree Gauss-Hermite quadrature filter, a new sampling rule named 5th-degree spherical radial cubature rule based on polynomials approximation and a new CKF with 5th-degree accuracy from this rule are deduced. The filter accuracy, sampling points, numerical stability are analyzed. The singular value decomposition algorithm is used to replace Cholesky decomposition for stronger stability. The experiment results demonstrate that the method is able to complete nonlinear alignment on big misalignment angles and has higher estimation accuracy than 3rd-degree CKF.