Research On Initial Alignment Algorithm Of SINS On Moving Base And Its Parallelization

Initial alignment is an indispensable process to Strapdown Inertial Navigation System (SINS). Only when the accurate initial attitude angles are acquired can SINS work normally during the navigation mode, including computing the attitude matrix, velocity and position. The initial alignment of SINS can be separated into two phase, i.e. coarse alignment and fine alignment. And the main characteristics of initial alignment are calculating time and calculating accuracy. However, the existent alignment methods cannot finish the alignment process swiftly and accurately when the SINS is located on a moving base. So, how to accelerate the alignment process and calculate accurately will be the subject of this paper.In terms of coarse alignment, a new initial attitude matrix computing method is proposed in the paper, in which the real time attitude matrix is changed into three different independent parts. The GPS velocity and position data are used to assist the alignment process, and are accumulated in different inertial coordinates together with IMU measurements in the form of integral operation. Then the transformation matrix of the inertial coordinates can be acquired and coarse alignment is completed. After that, an error model of SINS is founded in the fine alignment process, and Kalman filter is applied to estimate the attitude errors. In order to raise the calculating accuracy and upgrading frequency of initial alignment, the parallelization of alignment algorithms is studied in the paper, and different realization schemes are proposed for coarse alignment and fine alignment respectively. In the parallelization of fine alignment process, the faddeev algorithm and one dimensional systolic array are used to complete the matrix operation of Kalman filter.Several simulation tests are conducted to verify the accuracy and efficiency of the alignment algorithms and parallelization schemes. And a testing instrument is also erected, which is constituted of FPGA chip, GPS receiver and Inertial Measurement Unit (IMU). The final testing results show that the coarse alignment can be completed in12s with errors smaller than1°, and high accuracy can be acquired of fine alignment as well. The performance analysis and function verification of the parallelization of initial alignment confirm that the parallelization schemes for both coarse alignment and fine alignment are complete and effect.