Truncated Gaussian Fusion Estimation for Gyroscope Array With Bandwidth Maintenance Under Environmental Disturbance

Compared with the conflict between the bandwidth and accuracy based on Kalman filter (KF), the weighted fusion method can improve accuracy under environmental disturbances (EDs) while maintaining bandwidth. However, a challenge in this field is that the accuracy improvement of weighted fusion method has a limit (√n times). The primary reason is that the accuracy improvement of constant-weighted fusion method, as a multivariate linear function of weight coefficients, has a maximum value of √n. To break through this accuracy limitation, we prove the principle that the variance of Gaussian noise is equivalent to the variance of a truncated Gaussian EDs. A truncated Gaussian interval cluster fusion (TGICF) estimation is proposed based on this principle. To begin, the expectation is estimated using ordinary least squares through minimum variance fusion (MVF) measurement. The interval cluster is then calculated using noise, Eds, and expectation statistical information. Finally, samples outside the interval are adjusted into the interval by recalculating the weight coefficients. The experiments show that the gyroscope standard deviation (STD) is reduced from 20.53°/s to 3.23°/s under a vibration disturbance, and the frequency response is higher than -3 dB. These test results show that the accuracy is maintained while maintaining the bandwidth. Moreover, the accuracy is improved by 6.36, 2.65, 2.46, 2.63, and 2.72 times compared with original measurement, Ave, MVF, topical conflict measure (TCM) and scale-wise variance optimization (SVO) respectively, which means that accuracy limit is exceeded by 2.46 times. This method provides high robustness for gyroscopes used in high-end fields.