针对传感器在采集信号时混入不同的噪声，提出一种基于ICA-CEEMD小波阈值的组合去噪算法。该方法是对一维含噪信号进行剪切分段、平移和拼接，得到几个不同的含噪信号作为独立分量分析（ICA）的输入通道信号。通过ICA的盲源分离技术使得信号和噪声进行初步分离。再利用互补集合经验模态分解（CEEMD）对分离信号进行分解去噪，由于不同的高频和低频噪声，需要对分解的高阶和低阶固有模态函数（IMF）进行处理。对第一层和最后一层IMF利用3原则提取细节信息，进一步抑制模态混叠影响，重构去噪信号。最后，利用小波阈值对重构信号做去噪处理，提升去噪效果和性能指标。为验证该方法的有效性，进行了仿真和中北大学汾机实测实验，结果表明，该方法在去噪效果和性能指标上都优于小波软阈值去噪和基于CEEMD的小波阈值去噪方法，是一种有效的信号去噪新方法。
A method for sensor signal de-noising based on ICA-CEEMD wavelet threshold was proposed for the separation of different noise and signal. The method was to carry out the shear segmentation,translation and mosaic of one-dimensional noisy signals,and obtain several different noisy signals as the channel signal for independent component analysis (ICA).The signal and noise were separated by the blind source separation of ICA. The separation signal was decomposed by the complementary ensemble empirical mode decomposition (CEEMD). Due to different high frequency and low frequency noise,the high order and low order intrinsic mode components (IMF) of the decomposition needed to be processed. The way of 3 sigma principle was used to extract the detail information of the first layer of IMFs and the final layer of IMFs,and restrain the mode mixing effects and reconstruct the signal of de-noising. Then,the wavelet threshold was used to deal with the reconstructed signal,so as to improve the de-noising effect and the performance index. In order to verify the validity of the method,the simulation experiment and the Fenji test of North University of China were carried out. The results show that the proposed method is better than the wavelet soft threshold de-noising and wavelet threshold de-noising method based on CEEMD.
PDF(3025 KB)
