CAJ | 학술논문

针对传统小波包在诊断滚动轴承隐含故障中存在频率混叠、精度不高等问题，提出一种基于集成经验模态分解（ensemble empirical mode decomposition，EEMD）降噪与非抽样提升小波包相融合的故障诊断方法。首先利用 EEMD 方法分解原始故障信号得到多个本征模态函数（intrinsic mode function，IMF）分量，然后计算各个 IMF 分量与原始信号间的相关系数，并与设置的相关系数阈值相比较，将小于阈值的 IMF 分量视为伪分量予以剔除；对剩余的 IMF 分量采用峭度准则再次筛选最优 IMF 分量进行重构，进而实现降噪目的。为了避免传统小波包因采取抽样运算方式导致频率混叠情况，文中采用非抽样运算的提升小波包来分解降噪信号，并采用 Hilbert 变换进行包络解调分析得到滚动轴承的故障位置。仿真实验和滚动轴承内圈故障应用实例表明：采用 EEMD 分解原始故障信号，结合相关系数峭度准则，达到了很好的降噪效果；采用非抽样提升小波包比传统小波包具有更高的故障诊断精度，且不存在频率混叠问题。
침대전통소파포재진단곤동축승은함고장중존재빈솔혼첩、정도불고등문제，제출일충기우집성경험모태분해（ensemble empirical mode decomposition，EEMD）강조여비추양제승소파포상융합적고장진단방법。수선이용 EEMD 방법분해원시고장신호득도다개본정모태함수（intrinsic mode function，IMF）분량，연후계산각개 IMF 분량여원시신호간적상관계수，병여설치적상관계수역치상비교，장소우역치적 IMF 분량시위위분량여이척제；대잉여적 IMF 분량채용초도준칙재차사선최우 IMF 분량진행중구，진이실현강조목적。위료피면전통소파포인채취추양운산방식도치빈솔혼첩정황，문중채용비추양운산적제승소파포래분해강조신호，병채용 Hilbert 변환진행포락해조분석득도곤동축승적고장위치。방진실험화곤동축승내권고장응용실례표명：채용 EEMD 분해원시고장신호，결합상관계수초도준칙，체도료흔호적강조효과；채용비추양제승소파포비전통소파포구유경고적고장진단정도，차불존재빈솔혼첩문제。
Traditional wavelet packet in the implied fault diagnosis of rolling bearing exists some problems, such as frequency aliasing,the accuracy is not high,and so on.We propose a fault diagnosis method based on ensemble empirical mode decomposition (EEMD)de-noising and undecimated lifting scheme packet.Using EEMD method to decompose the original signals to obtain a lot of intrinsic mode function (IMF)components, calculated the correlation coefficients between each IMF component and the original signals,and compared with the threshold of correlation coefficients,if the correlation coefficients of IMF were less than the thresh-old,it would be deemed spurious IMF components and abandoned.The remaining IMF components were used kurtosis criterion to screen the optimal IMF components to reconstruct again,thus achieving the purpose of de-noising.In order to avoid the traditional wavelet packet produced frequency aliasing due to decimated opera-tion,we used undecimated lifting wavelet packet to decompose de-noising signals,and de-noising signals were demodulated with Hilbert transform to get rolling bearing fault location.The simulation experiment and the application examples of rolling bearing inner fault show that:using EEMD to decompose,combining correla-tion coefficient-kurtosis criterion,attains good de-noising;the undecimated lifting wavelet packet has higher fault diagnosis accuracy than the traditional wavelet packet,and do not exist the problem of frequency aliasing.