Abstract:In this paper a complementary ensemble empirical mode decomposition (CEEMD) de-noising algorithm for MEMS-gyro, is proposed to alleviative the drawbacks of the present empirical mode decomposition (EMD) forced de-noising method. The proposed method readily conditions signal distortion based on intrinsic mode function (IMF), as selected by composite evaluation; the results can be transferred into the linear combination of sample entropy (SE) and similarity. The filtering thresholdings are calculated by the signal noise parameters, and the relevant IMFs are filtered by those thresholdings. Simulation and test results show that the effect improves greatly as compared with the forced algorithm. In particular, the MEMS-gyro’s bias instability decreases by 76.4%, the MEMS-gyro’s rate ramp decreases by 62.3% and the MEMS-gyro’s RMSE decreases by 67.5%, after filtering by the proposed method. Furthermore, the filter improves the positioning accuracy of MEMS-IMU pedestrian navigation.