I. Introduction
The spacecraft attitude estimation system suffers from multisource interferences, including the unmodeled dynamic errors in the attitude estimation model, the measurement noise and structural vibration of the sensors inside the system, and torque disturbances resulting from the external working environment. These different types of interferences could seriously degrade the accuracy of spacecraft attitude estimation, which must be well tackled to provide desirable attitude estimation accuracy [1]–[3]. The classic attitude estimation algorithm has been applied to state estimation in actual engineering [4]. However, the algorithm still suffers from the degradation of filtering accuracy for the model error, internal noise, and external interferences mentioned above.