Fault diagnosis of rolling bearing based on wavelet threshold noise reduction EMD-AR spectrum analysis and extreme learning machine
-
摘要: 针对传统的滚动轴承故障诊断中,振动特征易受冗余噪声干扰,且不能对故障特征准确分类的问题,提出1种基于小波降噪、EMD-AR谱分析和ELM(极限学习机)的滚动轴承故障诊断的方法。对滚动轴承振动信号首先进行小波阈值降噪处理,随后将降噪后的一维信号进行EMD分解并提取其前6个IMF分量,将前6个IMF分量的AR谱累加得到降噪后振动信号的EMD-AR谱,可从谱中看出轴承不同的故障情况来作为先验诊断。最后提取降噪后信号的6个特征值作为样本,为避免实验的偶然性,建立基于K折交叉验证ELM分类诊断模型。诊断结果表明,该方法能对轴承故障情况进行清楚分类,分类精度最高可达100%,可对轴承故障诊断提供新的方法。Abstract: In the traditional rolling bearing fault diagnosis, vibration characteristics are easy to be interfered by redundant noise, and fault characteristics can not be accurately classified, a rolling bearing fault diagnosis method based on wavelet denoising, EMD-AR spectrum analysis and ELM (extreme learning Machine) was proposed. The vibration signals of rolling bearings are firstly de-noised by wavelet threshold, then the de-noised one-dimensional signals are decomposed by EMD and the first six IMF components are extracted. The AR spectra of the first six IMF components are accumulated to obtain the EMD-AR spectra of the de-noised vibration signals, from which different fault conditions of bearings can be seen as a prior diagnosis. Finally, six eigenvalues of the denoised signal were extracted as samples. In order to avoid the contingency of the experiment, the ELM classification diagnosis model based on K-fold cross-validation was established. The diagnosis results show that the method can clearly classify the bearing fault situation, the classification accuracy is up to 100%, and can provide a new method for the bearing fault diagnosis.
-
Key words:
- rolling bearing /
- fault diagnosis /
- threshold noise reduction /
- EMD-AR spectrum /
- ELM
-
表 1 组件参数表
组件名称 数量或参数 电动机 1.5 kW 转矩传感器 1 台 示功器 1 台 电控 1 套 加速度传感器 1 台 表 2 正常信号的特征值
工况 平均值 方差 最大值 最小值 峭度指标 波形因子
正常0.014 70 0.001 33 0.115 70 −0.082 25 3.459 07 1.307 16 0.014 14 0.001 75 0.125 12 −0.086 52 3.001 30 1.222 37 0.015 09 0.001 13 0.086 55 −0.055 00 2.457 99 1.164 05 0.010 37 0.001 59 0.102 04 −0.061 62 2.261 75 1.196 22 表 3 内套故障信号的特征值
工况 平均值 方差 最大值 最小值 峭度指标 波形因子 内套故障 0.002 97 0.007 91 0.521 42 −0.420 63 10.912 59 1.431 36 0.004 79 0.004 91 0.307 21 −0.327 43 7.378 57 1.382 04 0.005 76 0.006 12 0.388 34 −0.309 16 6.898 75 1.350 04 0.004 13 0.005 82 0.249 59 −0.292 07 3.789 02 1.281 29 表 4 滚动体故障信号的特征值
工况 平均值 方差 最大值 最小值 峭度指标 波形因子 滚动体
故障0.004 95 0.000 87 0.064 06 −0.054 60 2.093 01 1.182 57 0.005 36 0.001 08 0.107 09 −0.067 63 3.787 17 1.322 65 0.004 38 0.000 56 0.062 09 −0.044 65 2.919 65 1.230 67 0.005 22 0.000 65 0.069 73 −0.064 12 3.291 83 1.276 55 表 5 外圈故障信号的特征值
工况 平均值 方差 最大值 最小值 峭度指标 波形因子 外圈故障 0.003 86 0.027 43 0.750 23 −0.871 38 13.968 01 1.982 72 0.004 58 0.036 59 0.951 89 −1.197 16 19.221 75 2.159 15 0.003 93 0.026 56 1.260 57 −0.994 98 28.511 19 2.536 65 0.004 33 0.043 52 1.091 19 −1.202 59 16.612 26 2.099 04 -
[1] 杨洪涛. 样本熵改进小波包阈值去噪的轴承故障诊断[J]. 组合机床与自动化加工技术, 2020(1): 79-82,88. doi: 10.13462/j.cnki.mmtamt.2020.01.020 [2] 许立, 张宇, 汤武初, 等. 基于BP神经网络的圆锥滚子轴承故障诊断[J]. 组合机床与自动化加工技术, 2016(11): 105-107,111. doi: 10.13462/j.cnki.mmtamt.2016.11.028 [3] 邓飞跃, 丁浩, 郝如江. 基于多尺度特征融合残差神经网络的旋转机械故障诊断[J]. 振动与冲击, 2021, 40(24): 22-28,35. doi: 10.13465/j.cnki.jvs.2021.24.003 [4] 程尧, 陈丙炎, 张卫华, 等. 基于加权联合提升包络谱的轴箱轴承故障诊断[J/OL]. 西南交通大学学报: 1-9 [2022-06-12]. [5] 丁锋, 秦峰伟. 小波降噪及Hilbert变换在电机轴承故障诊断中的应用[J]. 电机与控制学报, 2017, 21(6): 89-95. [6] 贾峰, 李世豪, 沈建军, 等. 采用深度迁移学习与自适应加权的滚动轴承故障诊断[J]. 西安交通大学学报, 2022, 56(8): 1-10. [7] 何雷, 刘溯奇. EMD-AR谱分析和SVM的变速箱故障诊断[J]. 机械设计与制造, 2021, 369(11): 56-59,64. doi: 10.3969/j.issn.1001-3997.2021.11.013 [8] 姚峰林, 谢长开, 吕世宁, 等. 基于小波包变换和ELM的滚动轴承故障诊断研究[J]. 安全与环境学报, 2021, 21(6): 2466-2472. doi: 10.13637/j.issn.1009-6094.2020.0999