The accelerometer output signal is often superimposed on a large number of disturbing signals as a result of the characteristic of accelerometer and the external environment noise. Aiming at characteristic of accelerometer signal

focuses on the filtering of the accelerometer signal based on the wavelet threshold de-noising method. A vibration signal model is established with a real output signal superimposed

select optimal wavelet parameters and process the simulated signal by wavelet threshold denoising

the signal-to-noise ratio（ SNR） is increased from 4. 24 dB to 20. 45 dB

the root-mean-square error（ RMSE） is improved from 0. 051 to 0. 008 1. On this basis

real output signal of accelerometer is processed by denoising

and it shows that wavelet threshold denoising method works well on filtering acceleration signal.