Research On Fusion Structure Design And Multiplicative Noise Process For Satellite Attitude Determination System

The studying satellite attitude determination with high precision and high stability is the key technology of modern satellite, which has received increased domestic and overseas attention. For the sake of the requirement of satellites'high precision and high stability, it is achieved by adding the multi-source data integration or studying the more accurate of satellite model. The attitude determination system is studied systematically in this dissertation using the information fusion technology, optimal estimation theory, differential equations, the promotion of the kalman filtering technique, linear regression techniques and modern numerical methods for satellite attitude determination methods. Then novel feasible algorithms are proposed based on studying classical attitude determination algorithms.The new algorithms are simple but efficient, the computing cost is low but the attitude determination performs well.The main work is as follows:Firstly, we introduce the description of the satellite attitude and posture kinematics'model and the measurement principle and the measurement equation of several single sensors (star tracker / gyro / infrared horizon sensor / sun sensor) which is currently widely used.Secondly, in order to solve questions which are sensors'wrong data which increasing system reliability degradation and measurement information redundancy utilization, we give a federated filtering which can reduce computation and improve the accuracy of the satellite attitude determination based on the sun sensor/gyro, horizon sensor/gyro and star tracker/gyro. And a large number of simulation experiments validate the algorithm. On the same conditions, the attitude determination accuracy is greatly improved.Finally, based on the star-sensor and gyros pattern attitude determination system, we analyze the source of multiplicative noise and derive its form of measurement equation. Under the condition that the star-sensor's measurement model contains multiplicative noise, we can establish the star-sensor measurement equation with multiplicative noise. And we study the optimal filtering algorithm and the optimal smoothing algorithm in the linear minimum-variance. Computer simulations verify the feasibility and effectiveness of the algorithm. And we compare the two algorithms in different circumstances. The results showed that: when multiplicative noise is small, both are quite effective; when multiplicative noise is large, the accuracy of smoothing algorithm is significantly improved 40%.