Abstract:

According to the fact that when conventional cubature Kalman Filter (CKF) is adopted for strapdown inertial navigation system (SINS) initial alignment with large azimuth misalignment, the sampling points are proportion to the dimension of state vector, and calculation amount is large, reduced dimension CKF algorithm is proposed. Comparing with the conventional CKF algorithm, only the large azimuth misalignment angle is sampled in the discreted SINS nonlinear error model, and the third-degree spherical-radial cubature rule is used to calculate the posterior mean and covariance. The new approach reduces the sampling vector from 10 dimension to 1 dimension, and reduces the sampling points from 20 to 2, which reduces calculation amount. The simulation shows that new approach has the same alignment accuracy with conventional CKF algorithm, while the computational time is reduced to 1/3 of conventional CKF algorithm, which proves the practicality of the new approach.