中文摘要：

为了改进EMD中模态混叠的缺陷，提出一种基于SVD的模态混叠消除方法。SVD有两个特性，一是每个频率成分对应两个大小相当的奇异值；二是各频率成分经分解得到的奇异值大小与该频率成分的振幅呈正相关。该方法根据信号中主要频率成分构造出一组与原信号频率相等、幅值与原信号振幅成整倍数的正弦信号，并与原信号叠加后对叠加信号进行SVD分解，然后成对选取分解得到的奇异值重构出一组信号，依次减去前面加入的对应频率的正弦信号即可得到分解结果。实验表明，对于EMD分解模态混叠现象严重的信号，该方法能够进行有效消除模态混叠。

英文摘要：

An EMD mode aliasing elimination method is proposed based on SVD. The SVD has two features: (1) each frequency component corresponds to two sizeable singular values after decomposition; (2) these singular values have a positive correlation with the frequency. According to the major frequency components of the original signal, a series of sinusoidal signals are constructed whose frequencies are the same as those of the original signal but the amplitudes are integer times of those of the original signal. Then, these sinusoidal signals are superimposed with the original signal and the new signal is decomposed by SVD. Selecting the singular values in pairs from the decomposed new signal, another series of signals is reconstructed. The final decomposition results are achieved by subtracting the corresponding sinusoidal signals from the reconstructed signals. Experiments show that this method can eliminate the mode aliasing effectively even when the frequency components are close in the signal.