文献类型：期刊文献

中文题名：MEMS传感器随机误差Allan方差分析

英文题名：Allan variance analysis for the stochastic error of MEMS sensors

作者：高宗余[1];方建军[1];于丽杰[1]

第一作者：高宗余

通讯作者：Gao, Z.

机构：[1]北京联合大学自动化学院

第一机构：北京联合大学城市轨道交通与物流学院

年份：2011

卷号：32

期号：12

起止页码：2863-2868

中文期刊名：仪器仪表学报

外文期刊名：Chinese Journal of Scientific Instrument

收录：CSTPCD;;EI(收录号：20120314688635);Scopus(收录号：2-s2.0-84855721025);北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

基金：国家自然科学基金(50805004);模式识别与智能机器人系统学术创新团队项目(PHR201107149)资助项目

语种：中文

中文关键词：MEMS;Allan方差;随机噪声;功率谱密度;数学模型

外文关键词：MEMS; Allan variance ; stochastic noise ; power spectral density ; mathematical model

摘要：MEMS传感器中随机误差较大,有时会覆盖传感器中有用信号,提出采用Allan方差(Allan variance)方法对MEMS传感器实测数据进行分析,系统地分析了引起MEMS传感器误差的随机噪声种类及其来源和特性,确定其各项系数,根据系数获得其功率谱密度,根据功率谱密度分析法与Allan方差分析法获得对应各项随机误差的数学模型,然后以数学表达式的形式得到统一的数学模型,再与卡尔曼滤波相结合得到增强的卡尔曼滤波,最后通过车载实验对MEMS-INS/GPS各个姿态进行卡尔曼滤波与改进后卡尔曼滤波2种滤波方法的比较,实验结果表明新滤波方法能很好地提高微惯性系统各个姿态精度。
In some cases, large stochastic errors usually override effective signals in MEMS sensors. Allan variance approach as a common way in analyzing frequency stability in time domain has been proposed to analyze the data measured with MEMS sensors and describe the sources and characteristics of random noises that cause measurement errors of MEMS sensors. The coefficients of individual stochastic error are determined using Allan variance approach to obtain their power spectral densities. Meanwhile, the mathematical models of all stochastic errors are derived in combination with power spectral density analysis. A unified calibration mathematical model is obtained in the form of differential equation for multiple stochastic errors. An enhanced Kalman filter is designed to obtain better filtering effect through incorporating the unified calibration model into traditional MEMS-INS error equation. Finally, traditional and enhanced Kalman filters are applied to measure the poses of MEMS-INS/GPS in vehicle experiments and result shows that enhanced Kalman filter is superior to traditional one in improving the precision of micro inertial navigation.