Research On Several Key Technologies Of GFSINS

The common Strapdown Inertial Navigation System (SINS) isn't suitable for high overload and angular velocity, but the Gyroscope Free Strapdown Inertial Navigation System (GFSINS) can work validly in the high acceleration and high angular velocity conditions. GFSINS is widely researched and effective in recent years, but still can't meet the requirement of engineering application. To solve existing problems in this field, several key technologies are researched, such as GFSINS configuration, angular velocity calculation, attitude estimation and error calibration and compensation, and so on.Firstly, feasibility of system configuration is given: matrix J is left invertible. For the performance of system configuration, a kind of assessment standard based on Geometric Dilution of Precision (GDOP) is put forward and demonstrated. Considering the engineering realization, a ten-accelerometer configuration unit based on spinning projectile is designed. When one of accelerometers is trouble, four different arranging ways can guarantee system to work normally. At the same time, the judgment for GDOP and analysis for real condition show that the design is suitable for spinning projectile.Secondly, angular velocity calculation methods are studied for promoting navigation precision. Aiming at low precision of traditional algorithms, such as integral algorithm, extraction algorithm, differential algorithm, a lognormal algorithm with fast computation and high precision is given. Due to high calculation errors of traditional algebra algorithms, a three-layer BP network model is built to predict angular velocity, and the predicting precision is three times higher than that of lognormal algorithm. As eliminating the influence of random noises of accelerometer, a kind of H∞filter is designed. Simulation is complete under the different magnitude of noise, variety of noise and initial estimated value of state variable, and results show that H∞filter has a stronger robustness and stability than Kalman filter. Moreover, precision of angular velocity is improved by one order of magnitude higher than that of lognormal algorithm after H∞filter.Thirdly, attitude estimation for GFSINS is studied. Through simulation and comparison, the four samples rotation vector method has higher precision of attitude estimation than fourth-order Runge-Kutta method under cone rotation condition. Based on gravitational vector and magnetometers output, an attitude estimation method is put forward when carrier is in the static or constant motion state. To restrain error accumulation for different motion enviorment, a GFSINS/magnetometer integrated system is put forward. considering the cone error, the quaternion got by four samples rotation vector method and angular velocity obtained by ten accelerometers output, are chosen as state variable, cross-multiply output of angular velocity and measurement value of magnetometer are chosen as measurement variable, then the state equation and measurement equation are built. Moreover, EKF and SUKF filter are designed. Simulation results show that SUKF is superior to EKF, and attitude error can be inhibited validly after filtering.Finally, the calibration and compensation technologies for GFSINS are studied. Aiming at static and dynamic errors of accelerometer, all the error coefficients are calibrated using linear neural network based on error models, and the precision of angular velocity is improved by 1 order of magnitude after error compensation. An identification method for ten-accelerometer fixed position and direction errors is put forward, and 50 error coefficients can be calibrated at one time by system configuration completing three rolling under two rotating rates. After that, through using the method that firstly computes accelerometer output after compensation and then calculates angular velocity, the angular velocity precision is improved by more than one time.