基于CEEMD的小波软阈值和粗糙度惩罚平滑技术的联合信号去噪方法

为了有效去除采集信号中的噪声,基于MEMS水听器在采集信号时混入不同噪声的情况下,提出了一种基于CEEMD的小波软阈值和粗糙度惩罚平滑技术的联合信号去噪方法。CEEMD用于将一个含噪信号分解为几个固有模态(IMFS),然后把几个固有模态和原始信号作一个线性相关分析,分为相关性高的模态和相关性低的模态。将软阈值技术应用于相关性低的固有模态,并将粗糙度惩罚平滑技术应用于相关性高的固有模态,以提取尽可能多的信息,然后把处理后的新的固有模态重构形成去噪信号。分别在仿真和真实数据的基础上进行了实验,验证了方法的有效性。结果表明,联合信号去噪方法无论在去噪效果和性能指标上都优于基于CEEMD的小波软阈值的去噪方法和CEEMD的去噪方法,克服了经验模态方法和小波软阈值去噪的不足,为进一步分析与处理信号提供参考。

Denoising method based on CEEMD combine wavelet threshold and rough punishment

In order to denoise effectively during signal collection, and because MEMS hydrophone mixes different noises in signal acquisition, a combined signal de-noising method based on CEEMD soft threshold and roughness penalty is proposed. Signal containing noise by CEEMD method decomposes into several intrinsic mode (IMFS), and then a linear correlation analysis of intrinsic mode and original signal is conducted. Modals both with high dependency and low dependency are divided. We apply soft threshold technology to low inherent modals, and roughness of punishment is applied to the intrinsic modes of high correlation to extract as much information as possible, The intrinsic mode of denoising is reconstructed to form the new signal.In this paper, the method is used to test on the basis of simulation and real data.In order to verify the effectiveness of the method, experiments were carried out on the basis of simulation and real data respectively.The results show that this method is superior to both ceemd-based wavelet soft threshold de-noising method and CEEMD de-noising method in terms of de-noising effect and performance index. This method overcomes the shortcomings of empirical mode method and wavelet soft threshold denoising, and lays a foundation for further analysis of signal processing.