Research On Error Parameter Estimation Of Land High-Precision Ring Laser Cyroscope Strapdown Inertial Navigation System

With the development of ring laser gyroscope (RLG) produce technology in our country, research and utilization of the land high-precision RLG strapdown inertial navigation system (LHPRLG-SINS) have been attractive. To improve the performance of the LHPRLG-SINS, this thesis investigates theory and methods of error parameter estimation in the LHPRLG-SINS, including the inertial measurement unit (IMU) calibration, zero velocity update (ZUPT) and global observability analysis of the SINS. The main contents and contributions include the following aspects:1. The IMU analytic calibration error for the method based on a precise 3-axis turntable is derived, and an improved calibration scheme that can bate the turntable error is proposed. After establishing the IMU sensor model, we present an IMU calibration method based on a precise 3-axis turntable, which contains clockwise and counter-clockwise rotation tests along three axes, respectively, and a 24-position test. To analyze the calibration errors, this thesis sets up the turntable error model and the IMU measurement error model. The analytic calibration error is derived and proved by simulation and real tests, which describes the relationship between calibration error and the errors of turntable and inertial sensors. The calibration error implies the calibration precision limit using an arc-sec grade turntable. It is shown that the fiducial frame in the typical calibration scheme is nonorthogonal, and the frame for gyroscopes is not the same with that for accelerometers. Then we propose an improved calibration scheme, which bates the turntable error and sets up one common fiducial frame for gyroscopes and accelerometers. Real test results show that the new scheme improves IMU calibration precision.2. The multi-position norm-observed IMU calibration method is studied, a norm and vector-observed (NORVEC) calibration method with rotating excitation is proposed, and the measurement of turntable error using IMU is investigated. The NORVEC method overcomes some drawbacks of the norm-observed calibration methods before and weakens the dependence on turntable in calibration. Real tests show that the NORVEC method outperforms the traditional laboratory precise turntable-based calibration method. The optimal calibration scheme is investigated, and an approximately optimal calibration scheme is designed using the sensitivity function method and geometric method. After calibrated by the NORVEC method, an IMU can be used to measure the turntable error, which paves a new way for turntable error measurement.3. Two kinds of IMU systematic calibration methods are presented: fitting method and filtering method. Based on turntable information and inertial navigation information, respectively, two fitting systematic calibration methods are proposed. In calibration, the body frame should be restricted. The real reasons that there are 3 couples among 12 misalignment angles are indicated. Simulations show that the fitting systematic calibration methods are feasible, and the inertial navigation information-based fitting method is less affected by turntable error. For filtering systematic calibration method, we design two filters: two cases with and without lever arm are considered, respectively. The feasibility of the filtering method is verified by simulations. Real calibration tests were also performed for a LHPRLG-SINS using the filtering method. The size effect was considered. Test results show that the filtering systematic calibration method outperforms the traditional laboratory precise turntable-based calibration method.4. The ZUPT is used for a LHPRLG-SINS. The effect of IMU calibration errors on inertial navigation errors is analyzed and verified by simulations and tests. The attitude change complicates the effect of IMU calibration errors in the SINS, and as a result the navigation errors are not low-frequency oscillations. It worsens the performance of ZUPT. So IMU calibration is a key factor for ZUPT in the SINS. After calibrating IMU accurately, two ZUPT tests were performed for a LHPRLG-SINS. In both tests the horizon position errors are less than 7m (CEP) when the vehicle stopped about every 10 minutes, and less than 39m (CEP) when the vehicle stopped about every 20minutes.5. Global observability of the inertial navigation system is analyzed. From the observability definition and a global perspective, this thesis analyzes the observability of the inertial navigation system in general case, the static multi-position case and the case considering only constant biases, respectively. Some new conclusions are given and verified by simulations. Global observability analysis shows that there are fundamental weekpoints in the observability analysis based on the linearized system, and many observability conclusions derived from the linearization system are wrong in the global perspective.