A high-precision calibration method for nonlinear error coefficients of accelerometer components

With the continuous improvement of measurement accuracy requirements for inertial devices, how to accurately calibrate nonlinear error coefficients of accelerometer components has become an important factor affecting the accuracy of the inertial navigation system. Centrifuge calibration method can continuously provide a specific force greater than 1 g, which can fully excite the nonlinear errors of accelerometer components and is a commonly used method for calibrating nonlinear error coefficients. However, on the one hand, traditional centrifuge speeds are often selected based on empirical experience, lacking a scientific determination method. This can lead to a decrease in the calibration accuracy of nonlinear error coefficients. On the other hand, the inability to accurately model the highly complex and time-varying test errors during actual calibration further reduces the calibration accuracy. Therefore, a high-precision calibration method for nonlinear error coefficients is proposed. Firstly, by introducing G-optimal experimental design criterion to minimize the maximum scaled prediction variance of output prediction values, the optimal speed combination is designed to achieve the highest accuracy in estimating nonlinear error coefficients. Based on the idea of semi-parametric regression, system errors caused by calibration test errors are treated as parameters to be estimated, and a high-precision nonlinear error coefficient calibration model is established. Then the influence of calibration test errors is eliminated by estimating and compensating the system errors. Centrifuge calibration test results show compared with the traditional method, the ranges and standard deviations of the repeated calibration results of the proposed method are reduced by more than 80.37% and 63.01%. This indicates that the proposed method can effectively eliminate the influence of calibration test errors and achieve high-precision calibration of nonlinear error coefficients.