在线信号的快速经验模态分解方法

摘要

Traditional EMD has the problem of End Effects and slow decomposition,which can't be applied in real-time analysis for online signal.In order to solve this problem,a fast EMD for online signal was presented,according to the infinite long characteristics of online signal,firstly valid data of online signal was extracted,and then the signal envelopes were structured by fitting with LSSVR.The fast EMD effectively restrains the End Effects,and improves the quality and speed of EMD.The simulation results show that,the fast EMD is effective,which can realize the real-time decomposition for online signal,and has value in engineering application.%传统经验模态分解(EMD)中存在端点效应和分解速度慢的问题,无法适用于在线信号的实时分析.针对这一问题,提出了一种在线信号的快速EMD分解方法,利用在线信号“无限长”特点,首先提取在线信号有效数据,然后采用LSSVR拟合信号上、下包络线,不仅有效抑制了EMD分解的端点效应,而且大大提高了EMD分解的质量和速度.仿真结果表明:该方法快速有效,基本可以满足在线信号的实时快速分解,具有一定的工程应用价值.