基于改进EMD算法的信号滤波

为解决经典经验模态分解(empirical mode decomposition, EMD)滤波算法在低信噪比环境下滤波效果不佳的问题,提出了一种改进的EMD滤波算法。利用FFT对信号进行简单的频谱分析,若其中含有高频噪声,则对信号经EMD分解后得到的一阶本征模态函数 (intrinsic mode function, IMF)分量做剔除处理;若信号中含有白噪声及毛刺干扰,则向经典EMD滤波算法中添加变尺度因子,然后对信号进行EMD滤波,在算法最后一次迭代时再将一阶IMF剔除。仿真试验结果表明,改进的EMD滤波算法在低信噪比环境下有较小的均方误差值,滤波效果较好。

Signal filtering based on Improved Empirical Mode Decomposition

To solve the problem that the filtering effect of classical Empirical Mode Decomposition (EMD) algorithm wasn''t good in a low SNR environment, an improved EMD filtering algorithm was proposed. Signal spectrum was analyzed with FFT, if there was high-frequency noise in the signal, the first Intrinsic Mode Function (IMF) which was obtained by EMD algorithm was removed. And if there was white-noise or glitch in the signal, a variable factor was added to the classical EMD algorithm then filtered the signal using EMD, and the first IMF was removed at the last iteration of the algorithm. The simulation experimental results showed that the Mean Square Error (MSE) of the improved EMD algorithm was small and the filtering effect was good in the low SNR environment.