Nanosatellite Attitude Estimation with Uncertain Process and Measurement Noise Using Nontraditional Filtering

In this study the nontraditional attitude filtering algorithm integrates the Extended Kalman Filter (EKF) and Singular Value Decomposition (SVD) methods to calculate the attitude of a nanosatellite. Utilizing data from the magnetometer and Sun sensor, the SVD technique determines the attitude of the nanosatellite. The EKF is then provided with these attitude terms along with their error covariances, making the filter robust to changes in measurement noise. The Q (process noise covariance) adaption approach with multiple scale factors is suggested. This study shows that, the process noise bias and process noise increment type system changes will change the statistics of the EKF innovation. The theoretical basics of the Q-adaptive SVD-aided EKF with uncertain process noise mean and covariance are developed. Simulations are compared using the adaptive and non-adaptive versions of the nontraditional attitude filter in the presence of uncertain process and measurement noise.