经验模分解中的间歇性控制方法改进及分解精度控制

Improving the Intermittency Check Method for Empirical Mode Decomposition and Precision Control

提出了根据固有模式函数相邻零值点距离的间歇性控制准则,并提出通过多次间歇性控制来提高经验模分解滤波精度的方法。对合成信号分离的算例表明,与使用IMF相邻峰值距离的间歇性控制方法相比,使用这个准则分离出的信号更接近于原信号。本文还提出用IMF上下包络线平均值的最大幅值与IMF最大幅值的比作为EMD分解精度控制准则,这个准则更为直观和便于使用。

We propose the criteria for the intermittency check of the number of points between successive zerocrossings of an intrinsic mode function （IMF） and the method for enhancing signal-filtering precision of empirical mode decomposition （EMD） through repeated intermittency check. An instance of synthetic signal separation shows that the signals separated by using the criteria are more approximate to original signals compared with the intermittency check method that uses the number of points between successive extrema. Furthermore we propose more visual and easy-to-use criteria which use the ratio of the maximum of mean number between upper envelope and lower envelope to the maximum of the IMF for the control of the EMD precision.