摘要
针对齿轮箱故障诊断过程中,不可避免的存在噪声的问题,在自适应噪声完备集合经验模态分解CEEMDAN降噪方法的基础上,结合时频峰值滤波TFPF降噪方法,提出了CEEMDAN-TFPF联合降噪方法。首先对信号进行CEEMDAN分解得到模态分量,采用自相关和方差分析判断模态分量含噪程度,利用TFPF对含噪较多的分量进行降噪处理,将处理后的分量信号累加重构得到联合降噪方法的结果。通过仿真验证了该方法的可行性,利用实验验证了其可靠性,实验结果表明,所提出的方法能够对噪声进行有效抑制,且有效信号成分幅值不被削弱,降噪后信号信噪比提高,均方根误差减小,且能够有效地表现出故障的特征。
Addressing the inevitable noise in the process of gearbox fault diagnosis,we propose a CEEMDAN-TFPF joint noise reduction method that combines the traditional EMD noise reduction method and time-frequency peak filtering(TFPF).In this method,the signal undergoes decomposition via CEEMDAN to extract modal components.Autocorrelation and variance analysis determine the noise level of modal components,enabling the noise reduction process.The signal is then accumulated and reconstructed to achieve the joint noise reduction method result.The feasibility of the method is verified through simulation and its reliability through experiments.The results indicate that the method can effectively suppress the noise without weakening the amplitude of the effective signal component.Post noise reduction,the signal-to-noise ratio improves and the mean square root error decreases,effectively representing the fault characteristics.
作者
李民辉
李俊
刘雨
LI Minhui;LI Jun;LIU Yu(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China)
关键词
经验模态分解
时频峰值滤波
信号降噪
故障诊断
empirical mode decomposition
time-frequency peak filtering
signal denoising
fault diagnosis
作者简介
李民辉(1998-),男,江西赣州人,武汉理工大学机电工程学院硕士研究生.
