Accelerometer Configuration And Error Analysis For GFSINS

At present, the chief problem to make GFSINS (Gyro-Free Strap-down Inertial Navigation System) unavailable is the difficulties to obtain high-accuracy angular velocity, and the main factor to affect accuracy of angular velocity is the installation errors of accelerometers in GFSINS, but now all present accelerometer configurations are spatial axis configuration mode, and the advantage of this mode is that the progress is easy to design. However, it is not convenient to be engineering implemented and likely to cause large installation errors, thus reduce the navigation accuracy of GFSINS and affect its potential applications. This dissertation present a novel design of accelerometer coplanar configuration for GFSINS based on minimizing accelerometer installation error possibly, and this proposed coplanar configuration denotes that all accelerometers employed in GFSINS are situated on the same facet of the inertial measurement unit at the same elevation. The main task of dissertation is research on this coplanar configuration GFSINS and the mainly contents are as follows.1. The navigation principle discourse of GFSINS. On the base of introduction of common coordinate system in INS, the output expression of accelerometer of any location in GFSINS is derived. Then derive the principles formula of line movement and angular movement measurement with multiple linear accelerometers based on the output expression. And also infer the feasibility criterion of accelerometer configuration in GFSINS and the complete navigation equations of GFSINS.2. Research on accelerometer coplanar configuration in GFSINSA novel accelerometer coplanar configuration of GFSINS for missile is proposed in this article to minimize the installation error of accelerometer. All accelerometers in GFSINS are divided into "vertical accelerometer" and "parallel accelerometer" according to accelerometer installation features of coplanar configuration and the measurement characteristics of each category accelerometer are analyzed based on output expression of accelerometer. Next, on the basis of feasibility criterion above and the measurement characteristics of each category accelerometer, analyze the required conditions of accelerometers which must meet the feasibility criteria, that is, the options of "system accelerometer". Then for improving the accuracy of angular velocity, analyze the required conditions of "observation accelerometer" with the measurement characteristics of each category accelerometer. Lastly, for increasing system redundancy, analyze the redundancy scheme of accelerometers with the fault diagnosis theory of inertial sensors. And also for reducing the number of accelerometer possibly in GFSINS, a novel accelerometer redundancy scheme called "soft redundancy" achieved with software solutions which reuse the "observation accelerometer" as redundant accelerometer is proposed in this paper. On the basis of results above, a specific 12-accelerometer coplanar configuration is proposed and which rationality is analyzed. Then with the cost function used for designing redundancy scheme of inertial sensors in conventional gyro INS, the accelerometer redundancy schemes for various of accelerometer fault conditions is analyzed for the proposed 12-accelerometer coplanar configuration above. Last, the solving equations of specific force and angular velocity of 12-accelerometer coplanar configuration are obtained in case of that all accelerometers are in normal working condition.3. The nonlinear estimation of values of angular velocity in GFSINS. With the solving equations of angular velocity above, the discrete extended kalman state estimation formula of angular velocity is inferred. And the original angular velocity estimation model is deformed into another model with independent process noise and observation noise to deal with the correlated noise between process noise and observation noise of original estimation model. Last, consider further that the estimation error of EKF algorithm is ease divergence for the inaccurate model and statistical properties of noise, the adaptive fading kalman filter is introduced into the angular velocity estimation to suppress possible error divergence for various contingencies.4. The error analysis of GFSINS. The error sources of GFSINS are analyzed firstly, especially, the different error sources relative to conventional INS are compared. The complete error equations of GFSINS are inferred referring to the error equations of conventional INS. The main two types of errors of GFSINS——the measurement error and installation error of accelerometer are analyzed and obtain each error model. Last, the impact of two errors to accuracy of angular velocity is analyzed and compared with the foregoing simulation model of angular velocity.5. The error calibration and compensation of GFSINS.Study on the methods of calibration and compensation of accelerometer measurement error and installation error. And the accelerometer measurement error is calibrated firstly and then calibrates the installation error according to the difference of the two types of errors. At last, research on the real-time error compensation algorithm with the calibrated errors. There are mainly the following contents. According to the related literatures about the testing and calibration method for inertial components, the accelerometer static error model is calibrated with static rollover test in gravity field firstly and on this basis to calibrate the dynamic error model of accelerometer with three-axis turntable. With the difference of orientation error and position error in accelerometer installation errors, the accelerometer sensitive orientation error is calibrated firstly, which use the static rollover test in gravity field based on the calibration results of accelerometer measurement model. And for the 12-accelermeter coplanar configuration GFSINS above, two states of place for static rollover test in gravity field is designed to calibrate orientation errors of all accelerometers perfectly. The inclined plane turntable test in horizontal plane is designed to stimulate the position errors of accelerometer after the orientation error of accelerometer calibrated. For the installation characteristic of 12-accelerometer coplanar configuration, the three states of place in testing are used to calibrate the position errors of all accelerometers. According to the similarity of the two type's errors and also for not changing the foregoing estimation model of angular velocity, the combined error expression of two type's errors of accelerometer is inferred. Then an error compensation algorithm based on the nonlinear iterative least squares is proposed for real-time error compensation. And also this error compensation algorithm is verified to meet the accuracy requirements by a trajectory simulation.