基于自适应Kalman滤波的MEMS陀螺随机误差分析

发布时间：2018-01-19 06:24

本文关键词： MEMS陀螺随机误差自适应Kalman滤波时间序列分析自回归滑动平均 Allan方差　出处：《传感技术学报》2017年11期 　论文类型：期刊论文

【摘要】：针对某型MEMS陀螺随机误差较大、精度不高的问题,通过时间序列分析法,建立自回归滑动平均ARMA(AutoRegressive and Moving Average)模型,采用ARMA(2,1)模型将预处理后的MEMS陀螺随机误差进行建模。设计基于ARMA模型的经典Kalman滤波器。静态试验和恒定速率试验结果表明在经典Kalman滤波器作用下,静态试验下其均值与均方差下降32.62%和66.31%;恒定速率试验下,其均值有明显的降低,其均方差减小了一个数量级。针对经典Kalman滤波器不能解决振动试验中大振幅时滤波发散问题,提出一种新的自适应Kalman滤波法,通过寻找合适的标定因子s解决滤波发散问题。振动试验结果表明,当振幅为100°时,滤波后的均值和均方差分别下降8.25%和8.36%。
[Abstract]:Aiming at the problem of large random error and low precision of a MEMS gyroscope, the time series analysis method is adopted. An autoregressive moving average ARMA(AutoRegressive and Moving average model was established using ARMA(2. 1). The model models the preprocessed random error of MEMS gyroscope. The classical Kalman filter based on ARMA model is designed. The results of static test and constant rate test show that in the classical Kalman, With the filter. Under static test, the mean and mean square deviation decreased by 32.62% and 66.31; Under the constant rate test, the mean value is obviously reduced and the mean square deviation is reduced by an order of magnitude. The classical Kalman filter can not solve the problem of filtering divergence when the amplitude is large in the vibration test. A new adaptive Kalman filtering method is proposed to solve the filtering divergence problem by finding a suitable calibration factor s. The vibration test results show that the amplitude is 100 掳. The mean and mean square deviation after filtering decreased by 8.25% and 8.36 respectively.
【作者单位】：中北大学计算机与控制工程学院;江苏曙光光电有限公司;
【分类号】：TN96;V241.5
【正文快照】： 微机电系统(Micro-Electro-Mechanical System)惯性器件在无人机、精确制导武器、低成本惯导系统等领域得到大量应用。其中MEMS陀螺以其质量小、便于携带、易于安装、高可靠以及耐冲击等优势得到大规模使用。相比于激光陀螺、光纤陀螺、静电陀螺等,MEMS陀螺精度比较低,阻碍了M

【相似文献】