激光陀螺误差测试中非等精度问题的处理

Processing of variant-precision problem in error test of ring laser gyro

在环形激光陀螺的误差测试中,一般采用时域方差(或A llan方差)方法来分辨各误差项。取得不同采样时间下的方差(或A llan方差)σ2(τ),将其与采样时间τ进行回归分析,便可以分辨出各误差项。采样时间τ较大时,计算σ2(τ)所依据的数据点较少,因而算得的σ2(τ)可信度较低。然而,当剔除τ较大的σ2(τ)后,拟合所用的点数又较少,导致拟合结果可信度降低。针对这一矛盾,提出了用加权回归模型来处理激光陀螺误差分析中的非等精度问题。

Variance（or Allan variance） analysis method is widely used in the error test of ring laser gyro（RLG）. Various error sources are distinguished through regression of σ^2（τ） at different sample time. The dada points by which σ^2（τ） is calculated goes small when the sample time is very long, this makes bad precision of σ^2（τ）. If σ^2（τ） at long sample time are not used, the data for the regression points becomes less, and this brings lower precision of fitting result. Weighted regression model is proposed to processe variant-precision problem in error test of RLG.