介绍了Allan方差的基本定义,以及采用Allan方差对陀螺零偏数据进行处理所需的测试系统构成及测试要点。详细推导了采用Allan方差法对陀螺仪噪声进行估算的过程,描述了应用MATLAB进行数据处理平台建设的主要流程。最后运用Allan方差理论对MEMS陀螺信号进行了定量分析,得到了MEMS陀螺仪的量化噪声系数、角度随机游走系数、偏差不稳定性系数、速率随机游走系数和速率斜坡系数5个误差源系数,实验表明该方法能有效地辨识微机械陀螺的各项随机误差成分,可以正确评价陀螺仪的性能指标,为陀螺仪的设计改进提供了依据。 This paper introduces the basic definition of Allan variance and the test system constitution and test points that are needed to process gyro zero offset data by Allan variance. The process of estimating gyroscope noise by Allan variance method is deduced in detail, and the main flow of constructing data processing platform by using MATLAB is described. Finally, the Allan variance theory is used to quantitatively analyze the MEMS gyroscope signals. The coefficients of quantization noise, angle random walk coefficient, bias instability coefficient, rate random walk coefficient and rate ramp coefficient of the MEMS gyroscope are obtained. Experiments show that this method can effectively identify the random error components of the micromechanical gyroscope, and can correctly evaluate the performance of the gyroscope, which provides a basis for the design improvement of the gyroscope.