Satellite Attitude Determination Based On Hrg And Star Sensor

Hemispherical resonator gyros are widely applied in attitude determination system of high-performance spacecrafts by developed countries because of the merits of high reliability, high precision and long life. In this paper, the nonlinear filter theory and the testing technology in satellite attitude determination system combining hemispherical resonator gyros and star sensors are deeply researched. The development of the satellite attitude determination system based on domestic hemispherical resonator gyros is taken as the background. The main contents of the research are as following.The vibration law of hemispherical shell resonator, the variation law of rotational modality and the precession characteristics are analyzed from view of dynamics. The dynamic model of hemispherical shell resonator is approximately analyzed using analytic method. The effect of the structural parameters of hemispherical shell on resonant frequency and precession factor is investigated. The effects of the structural parameters such as thickness and diameter of hemispherical shell and pillar diameter on each order of natural frequency and sequence of vibration mode arrangement are investigated by the finite element method. Some reasonable suggestions are proposed for the structure design of the hemispherical shell resonator.The nonlinear compensation of the scale factor of the hemispherical resonator gyro is studied. The ARMA model of the random error of the hemispherical resonator gyro is identified. The output signal of the hemispherical resonator gyro is processed by the Kalman filter therefore the signal fluctuation decreases significantly.An algorithm based on extended Kalman filter for satellite attitude determination is presented, adopting modified Rodrigues Parameters in attitude representation. Results of the simulation experiment show that the convergence and the estimation precision of this algorithm are as good as the quaternion algorithm while the running time is much shorter. An algorithm based on extended Kalman filter for the horizontal attitude determination of the strapdown inertial attitude determination system is put forward under the failure condition of the star sensors. The simulation results indicate that the extended Kalman filter algorithm shows good convergence and good performance in restraining the accumulated integral error. The algorithm based on Unscented Kalman filter for satellite attitude determination is presented for the possibility of divergence of extended Kalman filter caused by local linear approximation under the condition of nonlinearity and big initial estimation error. The simulation results indicate that the Unscented Kalman filter algorithm shows better convergence and stability under bigger initial error angle, compared with the extended Kalman filter algorithm.An algorithm based on extended Kalman particle filter for satellite attitude determination is proposed. The simulation results indicate that the convergence rate of the extended Kalman particle filter algorithm is obviously faster than that of the standard particle filter algorithm and the Unscented Kalman filter algorithm. The attitude estimation error of the extended Kalman particle filter algorithm is only the half of that of the standard particle filter algorithm after convergence. In order to improve the filter performance, an algorithm for satellite attitude determination based on Unscented particle filter is proposed. The simulation results indicate that the convergence rate of the Unscented particle filter algorithm is fastest with least estimation error, compared with other algorithms.The prototype of the attitude determination system based on hemispherical resonator gyros and star sensors is designed and developed. During the course of research and development, the solutions were found for the critical problems such as hardware design, software design and testing of inertial components. The influence of the installation error to the redundant and the non-redundant inertial reference units is analyzed. The error model is set and the coefficients of the error model are identified with rational experimental method then the system installation error is eliminated by compensation. Simulations and experiments prove that all of the technical indexes have met the design requirements and the system is stable and reliable.