Even with bad geometry and constant rates, if gyro biases are independently known, the timetag error for a single sensor can be accurately estimated as long as its boresight is not too close to the spacecraft rotation axis.This is my HijackThis report if anyone could use it. Observability is much improved and the filter performs well when the rate is either varying or noncoplanar with the sensors, as during a slew. The EUVE geometry, though, is a degenerate case having coplanar sensors and rotation vector. The estimates are particularly sensitive to filter mistuning in this case. In the absence of attitude maneuvers, the state elements are highly correlated, and the state estimate is unreliable. Due to EUVE's sensor geometry, the observability of the state vector is severely limited when the spacecraft rotation rate is constant. The tests include cases with timetag errors on one or two sensors, both constant and time-varying, and with and without gyro bias errors. The flight data come from times when EUVE had a constant rotation rate, while the simulated data feature large angle attitude maneuvers. The examples use both simulated and actual flight data from the Extreme Ultraviolet Explorer (EUVE). The observability of the state vector is studied first through an examination of the algebraic observability condition and then through several examples with simulated star tracker timing errors. Thus, this filter is applicable to cases where the true timing errors are constant or slowly varying. The filter models the timetags as random walk processes: their expectation values propagate as constants and white noise contributes to their covariance. The state vector here consists of the attitude quaternion, timetag biases, and, optionally, gyro drift rate biases. Any timing errors in the observations can lead to attitude errors if either the spacecraft is rotating or the reference vectors themselves vary with time. Spacecraft attitude is determined by finding the mean rotation from a set of reference vectors in inertial space to the corresponding observed vectors in the body frame. This paper presents an extended Kalman filter for estimating attitude sensor timing errors.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |