基于ARMA建模的MEMS加速度计随机误差补偿方法

Random noise compensation method for MEMS accelerometer based on ARMA modeling

为提高MEMS加速度计测量精度,采用了一种基于自回归滑动平均(ARMA)模型和卡尔曼(Kalman)滤波的随机误差补偿方法。文中对预处理后的加速度计数据进行一阶差分,差分数据通过了平稳性分析,根据自相关和偏相关特性分析,确定随机误差适用模型,根据贝叶斯信息准则(BIC)确定模型阶数,从而确定随机误差模型。再通过Kalman滤波,实现对加速度计随机误差的滤波补偿,使得加速度计的零偏稳定性由0.5179mg降低为0.0528mg,指标提高了一个数量级,有效提高了加速度计的测量精度。

In order to improve the measurement accuracy of MEMS accelerometer,a random error compensation method based on ARMA model and Kalman filter is adopted.In this paper,the pre-processed accelerometer data is made into first-order difference.The difference data is analyzed by stationarity,and the applicable model of random error is determined according to the analysis of autocorrelation and partial correlation characteristics.The order of the model is determined according to the Bayesian information criterion(BIC),so as to determine the random error model.Then,Kalman filter was used to realize the filter compensation for the random error of the accelerometer,so that the zero-bias stability of the accelerometer was reduced from 0.5179mg to 0.0528mg,effectively improving the measurement accuracy of the accelerometer.