作者：刘自然, 胡毅伟, 石璞, 李谦, 尚坤

作者单位：河南工业大学机电工程学院, 河南 郑州 450007

关键词：滚动轴承;经验小波变换;频谱划分;特征提取

摘要：

针对振动信号的非线性、非平稳性和早期故障特征信号难以提取的特点，提出一种基于改进经验小波变换的故障特征提取方法。通过包络分析和对包络曲线进行阈值分割修整的方法来确定经验小波变换分解的模态数和频率边界，解决传统经验小波变换需要预先设置分解模态数和难以对信号频谱进行适当分割问题，以实现对振动信号故障信息更准确的描述。实验表明，该频谱分割方法能够有效检测信号最佳模态分解数，使得信号的频谱分割更为容易、可靠。相比传统EWT和EMD，改进经验小波变换的滚动轴承内圈、外圈Hilbert变换时频图对振动信号的故障相关特征描述更为清晰，在滚动轴承故障特征提取方面表现更为优越。

Fault feature extraction method of rolling bearing based on enhanced empirical wavelet transform
LIU Ziran, HU Yiwei, SHI Pu, LI Qian, SHANG Kun
School of Mechanical and Electrical Engineering, Henan University of Technology, Zhengzhou 450007, China
Abstract: Aiming at the characteristics of nonlinearities, non-stationary of vibration signals and the difficult to extract features of early fault characteristic signals, a fault feature extraction method based on enhanced empirical wavelet transform (EWT) is proposed. The modal number and frequency boundary of empirical wavelet transform decomposition are determined by envelope analysis and threshold segmentation of envelope curve. The problem that traditional empirical wavelet transform needs to pre-set decomposition modal number and is difficult to properly segment signal spectrum is solved in order to achieve a more accurate description of vibration signal fault information. Experiments show that the spectrum segmentation method can effectively detect the optimal mode decomposition number of the signal, making the spectrum segmentation of the signal easier and more reliable. Compared with traditional EWT and EMD, the Hilbert transform time-frequency diagram of inner and outer rings of rolling bearings based on enhanced empirical wavelet transform can describe the fault-related features of vibration signals more clearly, and it is better in extracting fault features of rolling bearings.
Keywords: rolling bearing;empirical wavelet transform (EWT);spectrum segmentation;feature extraction
2019, 45(10):10-15  收稿日期: 2018-12-07;收到修改稿日期: 2019-02-20
基金项目: 河南省自然科学基金(182300410234)
作者简介: 刘自然(1962-),男,河南信阳市人,教授,硕士,研究方向为动态测试技术、机电传动与控制技术
参考文献
[1] 吴印华, 徐琼燕, 李俊峰. 基于局部特征尺度分解排列熵和线性局部且空间排列的故障特征提取方法[J]. 机械设计与研究, 2017, 33(1):27-30, 34
[2] 张良, 张前图. 基于LCD模糊熵和流行学习的故障特征提取方法[J]. 机械强度, 2016, 38(2):225-230
[3] 宋金波, 王德平, 刘霞. 基于EMD瞬时功率谱熵的神经网络滚动轴承故障诊断[J]. 化工自动化及表, 2016, 43(8):793-796
[4] 李锋, 汤宝平, 陈法法. 基于线性局部切空间排列维数化简的故障诊断[J]. 振动与冲击, 2012, 31(13):36-40, 61
[5] SU Z Q, TANG B P, LIU Z R,et al. Multi-fault diagnosis for rotating machinery based on orthogonal supervised linear local tangent space alignment and least square support vectormachine[J]. Neuro-Computing, 2015, 157(1):208-222
[6] GILLES J. Empirical wavelet transform[J]. IEEE Transactions on Signals Processing, 2013, 61(16):3999-4010
[7] 刘自然, 陈仁权, 颜丙生, 等. 基于EWT和包络谱分析的轴承故障诊断研究[J]. 中国测试, 2018, 44(2):98-102
[8] 李志农, 朱明, 褚福磊, 等. 基于经验小波变换的机械故障诊断方法研究[J]. 仪器仪表学报, 2014, 35(11):2423-2432
[9] 向玲, 李媛媛. 经验小波变换在旋转机械故障诊断中的应用[J]. 动力工程学报, 2015, 35(12):975-981
[10] 祝文颖, 冯志鹏. 基于改进经验小波变换的行星齿轮箱故障诊断[J]. 仪器仪表学报, 2016, 37(10):2193-2201
[11] FRANCIS A, MURUGANANTHAM C. An adaptive denoising method using empirical wavelet transform[J]. International Journal of Computer Applications, 2015, 117(21):18-20