Periodic error detection and separation of magnetic levitation gyroscope signals based on continuous wavelet transform and singular spectrum analysis

The accuracy of the north azimuth measured using the magnetic levitation gyroscope (GAT) declines owing to the influence of the periodic errors of GAT signals induced by the systematic error of the gyro rotor system and the influence of external environment. To address this issue, this paper proposes a novel methodological strategy based on continuous wavelet transform (CWT) and singular spectrum analysis (SSA) to process GAT periodic errors and improve the accuracy of north-seeking. Firstly, CWT is used to process a large number of GAT signals to obtain the statistical characteristics of the periodic errors. Subsequently, the reconstructed components (RCs) of the GAT signals are obtained using SSA. After detecting and grouping the periodic terms contained in each RC using CWT, the periodic errors in the GAT signals are clearly separated. Finally, the effectiveness of this method was verified by comparing our north azimuths with those measured using the high-precision global navigation satellite system (GNSS) baseline. Our results indicated that the periodic errors in GAT signals can be accurately divided into the high frequency periodic error and the low frequency periodic error, and both periodic errors can be clearly detected and separated. After processing, the root mean square error of the GAT rotor currents and the absolute difference between the gyro and high-precision GNSS north azimuths were enhanced by 22.6% and 43.2%, respectively. The method presented in this paper to process periodic errors is suitable for use in the preprocessing of GAT signals.