动力调谐陀螺仪模型快速开环辨识

为了实现对动力调谐陀螺仪闭环测量系统的快速设计,建立了参数模型辨识系统,对激励信号设计、模型类结构选择、参数估计算法等问题进行研究.首先依据惯性元件的特点,设计了适宜于动力调谐陀螺仪测试的多正弦激励信号,然后基于外加输入自回归（ARX）模型,采用预报误差法辨识得到模型参数,最后针对某型号动力调谐陀螺仪设计实验进行验证,并与传统扫频法建模得到的结果进行对比.结果表明：模型辨识时间由2 000,s缩短为20,s,辨识拟合度由92.5%提高到95.3%.与传统扫频法相比,通过多正弦激励、ARX模型和预报误差法进行陀螺模型辨识可得到更高的辨识精度、抗噪声能力和辨识效率,为快速设计闭环系统提供了可靠的依据.

Fast Open Loop Model Identification of Dynamically Tuned Gyroscope

In order to complete the fast design of dynamically tuned gyroscope （DTG） closed-loop measurement sys- tem, this paper focuses on the establishment of a parameter model identification system, and presents a deep re- search on the type of excitation signal, model structure and parameter estimation algorithm. Firstly, multi-sine exci- tation signal is designed according to the peculiarity of DTG. Secondly, prediction error method （PEM） is introduced into model parameter estimation based on the auto regressive exogenous （ARX）model. Finally, a validation experi- ment is carried out for both parameter identification and comparison with the traditional swept frequency method. The results indicate that model testing time is shortened from 2 000 s to 20 s and the fitness degree is increased from 92.5% to 95.3%. The new modeling method possesses plenty of advantages in recognition accuracy, noise immunity and recognition efficiency, and provides a reliable basis for the design of closed-loop system.