Vibration Noise Suppression for Downhole Gyroscope Array Based on Dynamic Allan Covariance

Accurate attitude determination in measurement-while-drilling (MWD) systems is severely compromised by intense, time-varying, and spatially correlated vibration noise in downhole gyroscope arrays. Existing methods, including static Allan variance, single-sensor dynamic Allan variance (DAV), and conventional filters, fail to simultaneously address nonstationary noise and array spatial correlation, and rely on precalibrated noise statistics. This article proposes a novel vibration noise suppression method based on dynamic Allan covariance (DACOV), a dynamic extension of DAV from single-sensor variance domain to multisensor covariance domain. DACOV jointly characterizes time-varying nonstationary noise and intersensor spatial correlations in real time, which is integrated into an adaptive filtering framework to suppress vibration noise without prior noise information. Extensive simulations and vibration-table experiments demonstrate that the proposed method achieves 4.9 % reduction in angular rate root-mean-square error (RMSE) and 5.5 % improvement in bias stability under high-vibration conditions, outperforming conventional approaches for high-precision MWD applications.