28 / 2021-06-22 15:32:56

An improved initial alignment method using Kalman filtering of the vectorized K-matrix

Special Session > SS5 - Advanced Control, Navigation and Signal Processing Methods for Multiple Autonomous Unmanned Systems

In this paper, a Kalman filter based on the truncated vectorized K-matrix is proposed to improve the initial alignment process of SINS.  This filter makes up for the shortcoming that the Optimal-REQUEST algorithm estimates the scalar gain and performance index conservatively when calculating the K-matrix. The state space equation of the K-matrix is expressed as a matrix equation uniquely. Further, the state vector truncated by the K-matrix is used to establish the truncated state-space model. Meanwhile, the linear Kalman filter is employed to update the K-matrix. Then, the optimal quaternion can be acquired from the updated K-matrix to complete the initial alignment. Especially, the case of Gaussian white noise is considered so that it can reduce the impact of inertial device measurement error effectively during the initial alignment process. The simulation results imply that the proposed approach raises the accuracy and speed of initial alignment greatly in comparison with the traditional algorithm.