Volterra模型预测在EMD端点延拓中的应用

Application of Volterra Model Prediction to End Extension of Empirical Mode Decomposition

经验模态分解(empirical mode decomposition,简称EMD)的端点效应使得EMD分解结果产生严重失真,为了减小端点效应在分解过程中产生的影响,将混沌序列模型引入EMD,提出采用Volterra模型解决分解中产生的端点效应问题。论述了基于Volterra模型的数据延拓技术原理,即先对原始数据进行Volterra建模,然后利用该模型对数据进行延拓。该方法使端点处的延拓更加合理,从而使得三次样条曲线在端点处不会发生大的摆动,实现了准确的EMD分解。通过对仿真信号的研究表明,延拓抑制了分解的端点效应。把该技术应用于转子横向裂纹振动信号的EMD分解中,取得了良好效果。

This paper attempts to reduce the end effect resulting in a severe distortion of empirical mode decomposition(EMD).A chaotic sequence was introduced into the EMD end extension,and the Volterra model was adopted to solve the problem of end effect in EMD.The data extension technology based on the Volterra model was described.The method makes the end extension more reasonable,which prevents the cubic spline curve from swinging and so an accurate EMD is realized.The analyses of simulation signals show that this method can effectively control the end effect.