Research On Error Identification And Correction Of RLG Strapdown Inertial Navigation System

Choosing land high-precision RLG strapdown inertial navigation system as researchobject, the thesis aims to exploit several key techniques of error identification andcorrection, including the inertial measurement unit (IMU) calibration, single-axis rotationmodulation and zero velocity update (ZUPT). The main contents and contributions is asfollows:The mutual transformation relationship of quaternion, direction cosine matrix androtation vector is derived and the digital recursive navigation algorithm with rotation vectoris introduced. Based quaternion differential equation, the attitude error equation expressedby additive quaternion, multiplicative quaternion and rotation vector, respectively, isderived in detail and the equivalence of them three is proved.IMU calibration method based high-precision turntable is investigated. With IMUparameters model established, clockwise and counter-clockwise rotation test along3axesand24position test are designed to calibrate gyroscope and accelerometer. IMU calibrationformula based high-precision turntable is derived. After setting up the turntable error modeland IMU measurement error model, the thesis investigates calibration error caused by axisorthogonal error, angular position error of turntable and measurement error of gyro andaccelerometer. Two kinds of IMU systematic calibration methods are presented: fittingmethod and filtering method. Parameter fitting is made by establishing the relationshipbetween velocity error and IMU parameter error. The reduced sigma point scaled UKF isproposed, and the corresponding error model is set up. The simulation result conforms that,it yields the same calibration accuracy level as that of discrete calibration.In-depth and comprehensive study of single-axis rotation modulation of RLG SINS iscarried out. Auto-compensation principle of SINS single-axis rotation is analyzed. Theinfluence of IMU error on navigation error before and after rotation is investigated from theperspective of frequency domain. Innovative research is made to exploit optimal rotatingaxis, as well as the compensation effect of IMU drift, scale factor error and installation error.An axis selection scheme is proposed, which can modulate gyro and accelerometer drift inthree axial directions, leading to improving the navigation accuracy by two orders ofmagnitude. The influence of IMU drift error and axis position error on navigation accuracyof optimal rotating scheme is analyzed and simulation test is made. A lot of work of ZUPT of RLG SINS is completed in this thesis. Advantages andproblems of quadratic curve fitting is analyzed and solution to the problem is proposed. AKalman filtering method suitable for long parking time interval is presented, by which theZUPT experiment of RLG SINS is carried out. The positioning accuracy of50min is5.258m, which is improve by two orders of magnitude compared with pure INS navigation.Much attention is paid to the issue of zero velocity detection, the mathematicaldescription of which is made in detail. The detection problem is formalized as a binaryhypothesis testing problem. Three detectors are derived within the same general likelihoodratio test (LRT) framework, given the sensor model and signal model. Zero velocitydetection experiment using real IMU data is carried out, and the result shows that the GLRT,MV, ARE detector produce better performance than MAG detector.