A Method Based on VMD-JTFA-WP for Denoising of MEMS Gyroscope

The random error noise reduction of Micro-Electro-Mechanical System (MEMS) gyroscope can improve the accuracy of the gyroscope, and then affect the navigation accuracy of the unmanned system. A method of Variational Mode Decomposition (VMD) combined with wavelet packet Joint Time-Frequency Analysis (JTFA) is proposed to reduce the random error of gyroscope. Firstly, VMD is used to decompose the complex signal into its Inherent Mode Functions (IMF). Secondly, the correlation analysis in frequency domain and multi-scale permutation entropy in time domain are introduced as judgment criterion to divide complex signal into noisy IMFs, mixed IMFs, and signal IMFs. Next, the noisy IMFs are discarded directly, the mixed IMFs are denoised with wavelet packet, and the signal IMFs are retained. Finally, the denoised signal is reconstructed. The experiment of simulation signal and the static data of MEMS gyroscope is conducted. The Signal-to-Noise Ratio (SNR) of Bumps signal is improved by 1 dB compared with other methods, and the angle random walk (ARW) of gyroscope is reduced by 7.5^%, which shows the proposed method achieves a good denoising performance.