基于CEEMD的水下机器人MEMS陀螺降噪方法∗

摘要

MEMS-based gyroscope is vulnerable to many noises, especially high-frequency noises when executing underwater tasks,which is detrimental to a long-run system,requiring denoising data in real-time. Complementary Ensemble Empirical Mode Decomposition( CEEMD) is a novelty signal decomposition algorithm according to scales and sizes,and signal’s vibration gradually reduces with decomposition levels. And it can separate signals among dif-ferent frequencies. Taking MEMS gyroscope of underwater vehicle as a research object,this paper applies CEEMD to decompose gyroscope signals acquired during experiments in order to extract effective information. Meanwhile,Al-lan Variance is utilized to verify the effectiveness of CEEMD. Simulation results demonstrate that CEEMD has a good filtering effect on random noise and high-frequency signals.%MEMS陀螺仪工作时，容易受到各种噪声，尤其是高频噪声影响，不利于导航系统长时间工作，因此需要对数据实时去噪。互补集合经验模态分解( CEEMD)是一种按照自身尺度进行信号分解的算法，信号震荡随着分解级数逐渐减小，能够较好地分离高频和低频信号。以水下机器人MEMS陀螺仪为研究对象，根据水下实测数据，采用CEEMD分解陀螺信号，提取有效信息，并利用Allan方差验证CEEMD的有效性。仿真结果表明CEEMD对随机噪声、高频信号具有良好的降噪效果。