The Research On Attitude Determination Of Deep Sapce Explorer

Star sensors are considered the most accurate absolute attitude sensors in aerospace engineering, and they are becoming the preferable attitude sensors onboard most spacecraft these years. This paper presents and probes into several filtering schemes for the Deep Space explores attitude measurement system composed of star sensors and gyros.Under the stellar-inertial modes, two attitude determination algorithms are designed which use the extended Kalman filter. One of the algorithms is to linearized the state equation based on the optimal estimation. The other algorithms is to linearized the state equation based on the command quaternion. The quaternion is selected as global attitude parameter. By the definition of incremental error quaternion in the propagation equation of the full rank covariance matrix is derived, consequently the singularity of the covariance matrix caused by the constraint on the quaternion normalization is maintained.In order to improve the convergence properties, unscented Kalman filter(UKF) algorithm is derived under stellar-inertial modes. The UKF generates better estimates of the mean and the covariance of the states, leading to faster convergence.Two extended Kalman filters for star sensor based attitude and attitude rate determination are derived. In one methods, the attitude dynamics is used only to propagate the attitude rate. When high-bandwidth star sensor measurements are available, according to the Singer tracking model, the full angular acceleration is modeled as a first order Markov process while the use of the attitude dynamics is totally avoided.