中文摘要：

根据MEMS传感器中随机误差较大，有时会覆盖传感器中有用信号的问题，提出采用Allan方差方法对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，Allan方差分析方法的优势不仅在于能确定各类随机误差的误差特性，而且可以确定出其误差系数，由此可与相应的各种滤波方法相结合更好地实现其滤波效果。实验证明：Allan方差分析是对MEMS传感器随机误差进行研究的有效方法。

英文摘要：

Aiming at the problem of the stochastic error is large in MEMS sensors, and sometimes,it can cover the useful signals. Allan variance method is proposed to analyze experimental data of MEMS sensors. Varieties of random noises, source and characteristics are analyzed. The advantage of Allan variance analysis method is not only determine the characteristics of stochastic error, but also determine the error coefficients. So it can combine with the correspording various filtering methods to achieve better filtering effect. Experiments prove Allan variance analysis is an effective method to research stochastic errors of MEMS sensor.