基于CEEMDAN改进阈值滤波的微机电陀螺信号去噪模型

Signal de-noising model for MEMS gyro based on CEEMDAN improved threshold filtering

为了提高微机电系统(MEMS)陀螺信号的去噪效果,以自适应噪声的完备集成经验模态分解(CEEMDAN)方法为理论基础,并针对常规软阈值和硬阈值函数存在的不足,提出了一种基于改进阈值函数的CEEMDAN滤波去噪模型。该模型首先应用CEEMDAN方法将陀螺信号有效地分解为多个固有模态函数(IMF)分量;其次通过相关系数法判断噪声分量与有效分量的界限;进而对噪声分量进行阈值设置并使用改进阈值函数进行滤波处理;最后重构滤波处理后的噪声分量与有效分量以得到去噪后的信号。实际陀螺信号去噪试验结果显示:所提模型相对于CEEMDAN、集成经验模态分解(EEMD)、经验模态分解(EMD)强制去噪方法及小波分析方法,其信噪比提高了约3.9dB,均方根误差降低了约36%;所提模型相对于CEEMDAN结合软、硬阈值函数的去噪模型,均方根误差降低了30%以上。表明采用所提模型可以对MEMS陀螺输出信号进行有效去噪,提升去噪性能。

To improve the de-noising effect of MEMS gyro output signal based on the theory of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN),a CEEMDAN filtering de-noising model based on improved threshold function is proposed to overcome the shortcomings of conventional soft threshold and hard threshold functions.Firstly,the model uses CEEMDAN method to effectively decompose the gyro signal into multiple intrinsic mode function (IMF)components.Secondly,the boundary between noise components and effective components is determined by the correlation coefficient method.Then the threshold of each noise component is reasonably set up,and the improved threshold function is used for filtering.Finally,all noise components after filtering and all effective components are reconstructed to get de-noised signal.Through detailed de-noising analysis of actual MEMS gyro output signal,the results show that:1)the proposed model improves the SNR by a 3.9 dB and reduces the RMSE by at least 36%,compared with CEEMDAN,ensemble empirical mode decomposition (EEMD),empirical mode decomposition (EMD)forced de-noising method and wavelet analysis method;2)the RMSE by the proposed model is decreased by more than 30%,compared with the CEEMDAN de-noising model combined with soft and hard threshold function.The research results fully indicate that the proposed model can effectively de-noise the MEMS gyro output signal,and significantly improve the de-noising performance.