چكيده به لاتين
One of the important pillars of satellites is the attitude determination and control subsystem, which will have the task of stabilizing and orienting the satellite. Satellite attitude information is received through the sensors and then using the control rules and actuators, the required torque is provided for orienting the satellite in the desired condition. In this subsystem, the accuracy of the measurement information is very important because otherwise it causes the satellite to be in an inappropriate attitude. Due to the fact that the sensors are affected by systematic and random errors due to thermal effects, intense vibration during the launching process, instrument aging and noise caused by the environment or sensor itself, their accuracy is reduced. Therefore it is necessary to calibrate onboard them using offline or online methods.
In this study, the calibration of four sensors including a magnetometer, a star sensor, an analogue Sun sensor and a digital Sun sensor are investigated. First, errors sources are evaluated and based on these, a precise models of the sensors are extracted. Then the sensor output scenario is provided when mounted on a satellite and the satellite moves in a low orbit. In the following, in order to calibrate the magnetometer, a nonlinear two-stage least squares method and a two-stage Kalman filter method are proposed. The first method is based on an optimization problem and uses the combination of three algorithms including center solution, Levenberg Marquardt and linear least squares to find the internal parameters of the sensor including the bias, scale factor, non-orthogonal, and the parameters of the installation error. The second method is based on the combination of two algorithms including extended Kalman filter or unscented Kalman filter and linear Kalman filter. The advantage of these two methods is to find the internal parameters of the sensor with high accuracy. However, the online method is superior to the offline method due to low storage memory and the ability to calculate parameters at any one time, but the initial setup of the parameters is time-consuming. Througth the online methods, the combination of unscented Kalman filter and linear Kalman filter has a higher accuracy and more convergence speed, but its implementation is more difficult and time-consuming. The mentioned methods for calibrating the star tracker also are studied to obtain the internal parameters of the sensor including optical coefficients, distortion coefficients and the parameters of the installation error. Compared to previous methods, the advantage of these methods in calibrating the star tracker is to consider the relationship between optical coefficients and distortion coefficients, which increases the accuracy of calibration. Calibration of Digital Sun Sensors are also done in this thesis for the first time, and the methods mentioned above are also applied to these sensors. Finally, for the calibration of the analogue Sun sensor, linear least square method is also proposed for extraction of the bias and combinatorial Kalman filter for bias and installation errors.
Keywords: Attitude determination and control sub-system, on-board calibration, Magnetometer, star sensor, Analogue Sun sensor, Digital Sun sensor, Extended Kalman filter, Unscented Kalman filter, Linear Kalman filter.
