经验Bayes岭估计在GINS车载试验工具误差辨识中的应用

Application of Empirical Bayes Ridge Estimation in Instrument Error Identification of GINS Vehicle Test

在利用车载试验进行GINS工具误差辨识过程中,由于输入加速度很小使得系统存在严重的复共线性。应用传统的最小二乘方法会出现增大最小二乘估计量的方差、参数估计值不稳定、产生弃真错误等问题。本文引入经验Bayes岭估计方法来进行GINS车载试验工具误差辨识工作。仿真结果表明,和传统最小二乘方法相比,经验Bayes岭估计的辨识精度有所提高,并可克服系统存在的复共线性的影响。

During the identification of instrument error of gimbal inertial navigation system vehicle test, there is a serious multicollinearity exit in this system because the input acceleration is small. Application of the conventional least squares method can bring some questions, such as the variance may be enlarged and the parameter estimated value unstable and so on. In this paper, the empirical Bayes ridge estimation is proposed in the instrument error identification of GINS vehicle test. The simulation results showed that compared with the conventional least squares method, the empirical Bayes ridge estimation can enhance the precision of the parameter identification and can overcome the influence of the muhicollinearity.