Research On Integrated Attitude Determination Method Based On High Precision Attitude Sensor

The attitude information obtained by high precision attitude measurement equipment is of great significance to improve the optical remote sensing satellite’s image geometric quality and the target location accuracy.Due to the background of the increasing demand for satellite attitude accuracy,this thesis takes the attitude determination theory as the research basis,takes the multi-source data integrated processing as the core research thought,this thesis focuses on breaking through the integrated attitude determination algorithms of star sensor,gyro and angular displacement sensor,and realizes the research goal of improving satellite attitude accuracy and stability.The main research contents of this thesis include:(1)Combined with the attitude kinematics equation and attitude sensor measurement model,this thesis construct a linear system of integrated attitude determination.On the basis of analyzing the principle of extended kalman filter algorithm,an integrated attitude determination algorithm of star sensor and gyro based on forward-backward extended kalman filter is designed and implemented,which can effectively improve the utilization of attitude data.Compared with the traditional extended kalman filter,the accuracy of integrated attitude determination is improved.(2)The neural network theory is introduced into the field of integrated attitude determination,and an improved forward-backward adaptive extended kalman filter integrated attitude determination based on the limit learning machine is proposed,which can effectively suppress the star sensor measurement noise and improve the problem of systematic error in attitude determination model.The simulation results show that the proposed method has good generalization ability,and the accuracy of attitude determination is about 10% higher than that of forward-backward adaptive extended kalman filter in pitch and yaw direction,and its stability is better.(3)Taking the quaternion of star sensor as the measurement equation and considering the higher order error,the nonlinear attitude determination model is derived.By using the sampling point transfer state probability density to replace the linearization method,the integrated attitude determination algorithm of star sensor and gyro based on unscented kalman filter is designed and implemented.Simulation results show that the accuracy of the algorithm is better than that of the traditional extended kalman filter algorithm.(4)For strongly nonlinear non-gaussian systems,the idea of particle filter is introduced to design and implement the integrated attitude determination algorithm of star sensor and gyro based on particle filter.Simulation experiments verify that the algorithm is more accurate than the usual nonlinear filtering methods,such as unscented kalman filter algorithm.At the same time,compared with extended kalman filter,forward-backward extended kalman filter,unscented kalman filter and other algorithms,analyzed their characteristics and advantages,carried out different sampling rates,different precision of gyro and star sensor data integrated attitude determination performance simulation analysis,summed up the adaptability of the algorithm.(5)The simulation method of high frequency angular displacement data is designed and realized based on the theory of precise attitude determination.Combined with the characteristics of the angular displacement sensor model,the extended kalman filter,forward-backward extended kalman filter and unscented kalman filter algorithm are used to realize the integrated attitude determination of star sensor and angular displacement,which can effectively obtain high precision and high frequency attitude data.