Aiming at problem of disturbance of random noise of micro-electro-mechanical system（ MEMS）accelerometer on output signal

identification

estimation and modeling of accelerometer noise source and noise type are carried out

and determination of denoising method for error compensation are needed to improve precision of accelerometer. Allan variance analysis method is used to analyze on random noise of MEMS accelerometer

several main random noise which affect on performance of MEMS accelerometer are obtained. Mathematical modeling of accelerometer output data is carried out using auto-regressive and moving average（ ARMA） model

determine the used model and order by means of the final prediction error（ FPE） criteria. Design Kalman filtering algorithm

and use Kalman filter for noise reduction of accelerometer. The Allan variance method is used to analyze on filtering effect of Kalman algorithm. Experimental results show that the Kalman filtering can effectively reduce random noise of accelerometer.