基于EMD和ISA方法的轨道车辆动态应变信号去工频干扰研究

摘要

针对轨道车辆动态应变信号实际通道数量小于源通道,无法直接应用独立分量分析(ICA)方法降噪的问题,提出一种基于经验模态分解(EMD)和独立子空间分析(ISA)的去除工频干扰的方法.首先对信号进行EMD分解得到本征模态(IMF),再对结果进行ICA分析,得到独立分量,然后对独立分量进行层次聚类将原信号分解到不同的子空间中,从而达到分离噪声的目的.实验结果表明该方法降噪效果优于传统滤波算法,处理后的信号损伤值高于带阻滤波值处理结果,使寿命评估结果偏向于安全,证实了将其应用于去除轨道车辆动态应变信号中工频干扰的可行性.%As for the dynamic strain signal of the track vehicle,the number of acquisition channels is smaller than that of source channels and it is difficult to apply Independent Component Analysis (ICA) algorithm for noise reduction.To solve this problem,a method for the removal of power frequency interference is proposed,which is based on Empirical Mode Decomposition (EMD) and Independent Subspace Algorithm (ISA).First,the signal is decomposed into a series of Intrinsic Mode Functions (IMF) through EMD.The results are analyzed through ICA algorithm to get the independent components.Then,the original signal is decomposed into different sub spaces by clustering ICs using hierarchical clustering algorithm so that the purpose of separating noise can be achieved.Experimental results show that the proposed method is superior to the traditional filtering algorithm in noise reduction,and the processed value of the signal damage is higher than that of the band stop filter,which makes the life assessment results appear safer.Applying this method to removing the power frequency interference in the dynamic strain signals of track vehicles is thus proved feasible.