水下机器人陀螺仪降噪优化设计研究

De-Noising Analysis of Gyroscope Based on CEEMD Reconstruction Algorithm

针对水下机器人MEMS陀螺仪存在噪声,为提高导航精度,解决高、低频噪声难以分辨和剔除的问题,提出了将互补集合经验模态分解(Complementary Ensemble Empirical Mode Decomposition,CEEMD)与相关性理论相结合的方法优化降噪算法。首先采用CEEMD算法分解MEMS陀螺仪信号,获得一系列本征固有函数(Intrinsic Mode Function,IMFs),以各IMF与原始信号的互相关系数为依据剔除虚假分量,完成对陀螺仪的降噪处理。最后利用阿伦(Allan)方差进行效果验证。仿真结果表明,与传统的直接去除高频的IMF分量法相比,改进后的CEEMD重构方法最大限度地保存了有效信息,显著提高了陀螺仪的输出精度。

In order to reduce the noise and improve the navigation accuracy of the underwater robot, to solve the problem of distinguishing high frequency and low frequency noise, a combination of Complementary Ensemble Empirical Mode Decomposition （CEEMD） and the correlation theory is proposed. Based on the data collected from the gyroscope, the CEEMD algorithm is used to decompose the gyro signal, extract the effective information and calculate the mutual correlation coefficient between each IMF and the original signal, as well as eliminate the false components. Finally, the Allan variance is adopted to verify the results, and it turns out that, compared with the traditional CEEMD denoising method, the improved CEEMD reconstruction method can save the most effective information and improve the output precision of the gyroscope.