基于小波方差的MEMS IMU随机误差模型间接估计方法  

Indirect estimation method for random error models of MEMS IMU based on wavelet variance

作　　者：刘菲[1] 任章[1] 李清东[1] 

机构地区：[1]北京航空航天大学自动化科学与工程学院,北京100083

出　　处：《中国惯性技术学报》2016年第1期77-82,共6页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金重点项目(61333011)

摘　　要：MEMS IMU(Mirco Electro Mechanical System Inertial Measurement Unit)微机电惯性测量模块广泛应用于组合导航系统,其中MEMS IMU随机误差模型的准确性对导航精度有着重要的影响。针对Allan方差法在估计随机误差模型方面的不足,研究了间接估计方法。该方法以Daubechies离散小波变换与间接推断原理为基础,根据小波系数的零均值平稳特征,对分解尺度进行确定,将小波方差作为间接估计辅助参数,分析了最优估计准则的渐近一致性,最后使用高斯牛顿法对估计结果进行校正,获得满足渐近一致性的随机误差模型参数估计结果。仿真结果表明,间接估计方法提高了随机误差模型的估计精度,其中一阶马尔科夫过程的相关时间估计精度提高了12.383%,解决了一阶马尔科夫过程模型的准确估计问题。通过试验结果分析,进一步证明了以上结论。Recently, MEMS IMU has been widely used in the integrated navigation system, and the accuracy of random error models has important effects on the navigation information. Regarding the disadvantages of the Allan variance method in parameter estimation, the indirect estimation was proposed. On the basis of Daubechies discrete wavelet transform and indirect inference theory, wavelet decomposition scale was determined by wavelet coefficient series properties. Then wavelet variance was adopted as the auxiliary parameter of indirect estimation, and an optimal criterion possessing progressive consistency was researched. Finally, Gauss-Newton method was utilized to calibrate the estimated results of random error model parameters which satisfied asymptotic consistency. Simulation indicated that, compared with Allan variance method, the indirect estimation method improved the accuracy of parameter estimation, and the accuracy of a first-order Markov random process correlation time was improved 12.383%. So the indirect estimation method effectively resolved the issue concerning accurate parameter estimation of a first-order Markov process model. By analyzing test results, the conclusions were verified further.

关 键 词：MEMS IMU 随机误差模型 Daubechies离散小波变换 小波方差 间接推断 

分 类 号：U666.1[交通运输工程—船舶及航道工程]