基于EMD的水轮机空化声发射信号阈值降噪方法

摘要

针对随机噪声对水轮机空化声发射信号的影响,提出了基于经验模态分解(EMD)的水轮机空化声发射信号阈值降噪方法:首先利用EMD将水轮机空化声发射信号分解得到多个本征模态分量(IMF),然后结合给定阈值规则对前半部分IMF分量进行降噪,将降噪后的IMF分量与未处理的IMF分量重构,得到降噪后的声发射信号.采用该方法分别对仿真信号和水轮机空化试验过程中采集到的声发射信号进行了降噪处理,并与小波阈值降噪方法、小波包闽值降噪方法及传统EMD降噪方法进行了对比.结果表明:EMD阈值降噪方法有效结合了EMD自适应分解的特性和阈值降噪的良好性能优点,对水轮机空化声发射信号的降噪效果优于其他3种方法.%Aiming at the influence of random noises on the acoustic emission signals in the cavitation condition monitoring of hydraulic turbines,a threshold denoising method was proposed based on empirical mode decomposition (EMD) for acoustic emission signals from hydraulic turbine cavitation.Firstly,EMD was used to decompose the acoustic emission signals to some intrinsic mode functions (IMFs).Then the first half of IMFs were denoised according to the given threshold rules.Both the denoised IMFs and the unprocessed ones were reconstructed to get the denoised acoustic emission signals.This method was implemented on the simulated signals and those collected in a cavitation experiment,and the denoising effects were compared with those obtained by the wavelet threshold denoising method,the wavelet packet threshold denoising method and the conventional EMD denoising method.Results show that the method proposed is better than other three algorithms in the noise reduction of acoustic emission signals from hydraulic turbine cavitation,which combines the self-adaptive decomposition characteristics of EMD and the advantages of threshold denoising method with good performance.