Research On Strapdown Inertial Navigation System Errors Self-Calibration

The methods utilizing the navigation computer of SINS (Strap-down inertia navigation system) to estimate its errors could be called SINS self-calibration technique. Because the initial misalignment and the errors of the inertial sensors are the two most important factors affecting the performance of SINS, the thesis mainly deals with the initial alignment technique and the calibration technique based on the self-calibration technique.Toward stationary alignment, the state-space error model of SINS is analyzed firstly. The thesis then simplifies it in the case of stationary base, and presents the integral alignment method.Toward transfer alignment(TA), the thesis mainly deals with the velocity-plus-attitude matching scheme based on the observability analysis. Firstly, the error mathematic models of several matching scheme in TA are stated. According to linear system theory, the observability degrees of the state variables are analyzed in the maneuvers of constant velocity and constant velocity plus rolling when the velocity-plus-attitude matching scheme is used. The thesis finally shows the reason that the velocity-plus-attitude matching scheme can perform quick alignment in the maneuver of constant velocity plus rolling.Toward calibration, the thesis presents the calibration method of the inertial sensors' errors based on TA technique. Based on the error mathematic models of several matching scheme in TA, the influences of gyro constant drift rate, gyro scale factor error, gyro misalignment and accelerator constant bias are considered. Then the thesis designs two systematic calibration approaches to calibrate these errors. The corresponding calibration path is also designed depending on the degrees of the observability. In view of the unsatisfactory estimation results of the horizontal accelerator constant bias, an improved approach is presented, which rotates the system according to the optimum path under stationary alignment.The results of experiment and simulation show that these approaches are feasible, efficient, and valuable in practice.