Research on random error analysis and noise reduction method of MEMS gyroscope

Abstract:
Aiming at the low measurement accuracy caused by random error and outliers in micro electro mechanical system (MEMS) gyroscopes, an adaptive Kalman filtering and wavelet analysis based on autoregressive integrated moving average model (ARIMA) is proposed to eliminate outlier noise. Allan variance analysis method is used to analyze the main random noise of MEMS gyroscope in this scheme. ARIMA was used to model the output signal of MEMS gyroscope. The adaptive Kalman filter equation is established to suppress the random error and outliers of MEMS gyroscope. Wavelet analysis is used to denoise the filtered signal. The experimental results show that the quantization noise coefficients of x, y and z decrease by 93.1%, 94.3% and 88.2%, respectively, compared with before denoising. The angle random walk noise of x, y and z decreases by 98.8%, 98.3% and 97.4% respectively. The zero bias instability of x, y and z decreases by 15.6%, 94.6% and 90.7% respectively, which proves the effectiveness of the proposed scheme in improving the measurement accuracy of MEMS gyroscope.