多级降噪联合特征增强的轴承故障诊断  

Multi-stage denoising combined with feature-enhanced rolling bearing fault diagnosis

作　　者：廖运虎 纪国宜[1] LIAO Yunhu;JI Guoyi(School of Aeronautics,Nanjing University of Aeronautics and Astronautics,Nanjing 210000,China)

机构地区：[1]南京航空航天大学航空学院,南京210000

出　　处：《振动与冲击》2025年第8期199-208,共10页Journal of Vibration and Shock

摘　　要：对于强噪声背景下,滚动轴承早期故障特征难以提取的问题,提出以改进奇异值分解(improved singular value decomposition,ISVD)联合改进小波分解的多级降噪为预处理,以参数自适应多点最优最小熵解卷积(multipoint optimal minimum entropy deconvolution adjusted,MOMEDA)特征增强为后处理的新方法。先是针对奇异值分解难以选择奇异值的问题,提出一种ISVD降噪方法,避免了奇异值的选取;针对软、硬阈值小波降噪的缺陷,提出一种新的小波降噪方法。针对MOMEDA中多点峭度谱求解周期时易受噪声干扰问题,首先利用多级降噪对信号进行降噪预处理,再利用新的周期指标多点包络峭度谱识别周期。通过仿真以及试验验证了该方法的有效性和优越性。Since it is difficult to extract the early fault features of rolling bearing in the background of strong noise,a new method was proposed in this paper.Pre-processing was performe dthrough multi-stage denoising that combining the improved singular value decomposition(ISVD)with the improved wavelet decomposition,and post-processing was performed through parameter adaptive multipoint optimal minimum entropy deconvolution adjusted(MOMEDA).Aim at singular value decomposition denoising method is hard to select singular value,an ISVD denoising method was proposed to avoid the selection of singular value firstly.Aim at the defects of soft and hard threshold wavelet denoising,a new wavelet denoising method was proposed secondly.In view of the problem of multi-point kurtosis spectrum is sensitive to noise in solving the cycle,multi-stage denoising was used to pre-process the signal,and then the new cycle index multi-point envelope kurtosis spectrum was used to identify the cycle.The effectiveness and superiority of the proposed method was verified by simulation and experiment.

关 键 词：多级降噪 改进奇异值分解(ISVD) 改进小波分解 多点包络峭度谱 强噪声 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]