Research Of Spacecraft Attitude Determination Algorithm Based On Narrow FOV Star Sensor And Gyroscope

The development and application of navigation technology make it suitable for a variety of workplaces,but a single navigation system is more or less inadequate.Therefore,we make up for the deficiencies of each single navigation system by means of integrated navigation and realize the optimization of the combination system process.In a variety of integrated navigation systems,can not be ignored is the combination of inertial navigation and astronomical navigation.Among the most commonly used attitude sensors include gyroscopes,accelerometers,star sensors.The combination of a gyroscope and a star sensor can overcome their respective drawbacks and is a common combination.Compared with the wide field star sensor,the narrow field star sensor has advantages in terms of volume and weight,and the narrow field star sensor is less susceptible to the interference of the external environment under the interference environment of mid-low altitude flight altitude.In this dissertation,a narrow-field star sensor and a gyroscope are selected to determine the system model based on the National Natural Science Foundation of China(61573113)and the Harbin Special Prize for Innovative Talents(2014RFXXJ074),and the attitude filtering algorithm suitable for the narrow field star sensor the study.Specific work as follows:According to gyroscope and star sensor,we establish the model.Since the actual attitude estimation system generally belongs to a nonlinear system,the orbital dynamics equations of the quaternion are discretized directly to obtain the discrete equation of state,while the noise term of the equation of state has the coupling noise term selected by the quaternion.The coupled noise term is analyzed and the mean and variance of coupled noise term are calculated.The performance and applicable conditions of different filtering algorithms of nonlinear system are analyzed.The ambiguity problem exists when the quaternion is used to describe the moving state of objects.According to the multiplicative property of quaternion,the perturbed sampling points are selected to estimate the state and error variance.Based on the extended kalman filtering algorithm,unscented kalman filtering algorithm and the structure of volumetric kalman filtering algorithm,the pose estimation is carried out using multiplicative EKF algorithm,quaternion-based UKF algorithm and quaternion-based CKF algorithm respectively,and the simulation analysis The performance and stability of each algorithm show that the quaternion-based UKF and CKF filtering algorithms are more suitable for large initial error conditions.Attitude algorithm based on the combination of star sensor and gyroscope in narow field of view.Due to the limited number of stars captured by a narrow field of view,the minimum requirements of recognition matching may not be satisfied,and the loss of recognition accuracy caused by the loss of information may affect the accuracy of pose estimation.In order to solve the problem of star point loss,the expansion of the navigation star catalog adds more dark stars and increases the time of star identification,which delays the attitude estimation.Aiming at the uncertainty model with loss of measurement,a robust recursive filtering algorithm is proposed to solve the problem of star sensor information loss.A robust extended kalman filter(REKF)algorithm is proposed for the uncertain nonlinear system model with delay measurement.The problem of star sensor delay is solved.The complexity of the system makes it difficult for us to solve the exact value of the variance matrix.According to the principle of minimum mean square error criterion,we can solve the minimum upper bound of each variance and design the filtering gain by the minimum upper bound.The stability of the proposed REKF algorithm is analyzed to determine the mean square of the estimation error of the REKF algorithm.Theoretically,the feasibility of the REKF algorithm is illustrated.Through the simulation analysis,the validity of the algorithm under different conditions is verified.