Abstract:

      To reduce the amount of calculation and solve the instability problem in the strapdown inertial navigation system(SINS) initial alignment of large azimuth misalignment, this paper proposes a modified square root unscented Kalman filter (SRUKF). The filter uses the spherical simplex unscented transformation to reduce the number of sigma points to speed calculation without sacrificing the filter precision. To reduce rounding errors and improve stability, the covariance matrix is replaced with a new matrix whose entries are square roots of the covariance matrix. Simulation results show that the spherical simplex SRUKF algorithm can significantly reduce the computation load and provide better filtering performance with accurate alignment.

Key words: spherical simplex transformation , square-root filter , initial alignment