Adaptive Kalman Filter Enhanced With Spectrum Analysis for Wide-Bandwidth Angular Velocity Estimation Fusion

The optoelectronic inertial stabilization platform is widely used in the fields of astronomical observation, monitoring and search, quantum communication and other fields, which all need strong vibration suppression and tracking capability. To realize high-precision angular measurement throughout the wide bandwidth for inertial stabilization platform, the fusion algorithm is studied to fuse the signals of magnetohydrodynamic (MHD) and microelectromechanical (MEMS) gyroscope. The basic adaptive Kalman filter experiences signal-to-noise ratio and fusion-frequency jitter problems because of the measurement transfer matrix deviation, making it unsuitable for quick dynamic systems. This paper proposes an adaptive Kalman filter enhanced with spectrum analysis by connecting the measurement covariance with the signal frequencies. This method divides the trace of the modified measurement covariance into three parts according to frequency domain characteristics. Based on the frequency analysis, the filter output mainly depended on the MEMS gyroscope at low frequencies, fusion results of the two gyroscopes in the intermediate frequency-domain, and the MHD gyroscope at high frequencies. This theory is verified in the swept frequency experiment and comparative experiments, including multi-harmonic sinusoidal and step response tests compared with combing filter and closed-loop filter. The tests results show that the fusion signal is identical with the actual signal throughout the measurement bandwidth, and the fusion signal-to-noise ratio is improved. The frequency characteristics and noise level of the fusion algorithm satisfy the requirements of optoelectronic inertial stabilization vibration measurement.