Bearing Remaining Life Prediction Method Coupling with Acoustic Emission and Convolutional Neural Network
-
摘要: 根据声发射信号具有对早期损伤敏感性高和频带宽等特点,提出一种结合声发射信号和卷积神经网络的方法,实现滚动轴承的RUL预测。该轴承RUL预测方法主要包括:对原始信号的分频段滤波和特征值提取得到高维特征集;将高维特征集组合成二维神经元作为卷积神经网络的输入,并构建和训练网络以达到预测剩余寿命的目的。通过从实验中得到的数据验证了该预测方法的可行性,并且具有较高的准确性。结合使用卷积神经网络后不但解决了特征值数量大和如何合理利用高维特征问题,而且还得到了较好的RUL预测效果。Abstract: According to the characteristics of acoustic emission signal with high sensitivity to early damage and frequency bandwidth, a method coupling with acoustic emission signal and convolutional neural network was proposed to realize the RUL prediction of rolling bearings. The bearing RUL prediction method mainly includes: sub-band filtering and feature value extraction of the original signal to obtain a high-dimensional feature set; combining with the high-dimensional feature set into a two-dimensional neuron as the input of a convolutional neural network, and constructing and training the network to achieve the prediction of remaining life. The feasibility of the prediction method is verified by using the experimental, and it has high accuracy. The coupling use of convolutional neural networks not only solves the problem of large number of eigenvalues and how to reasonably use high-dimensional features, but also obtains the better RUL prediction results.
-
表 1 时域特征
Table 1. Time-domain features
序号 特征指标 序号 特征指标 1 均值 9 均方根值 2 最大值 10 峰峰值 3 最小值 11 波形指标 4 歪度 12 脉冲指标 5 峭度 13 峭度指标 6 方差 14 峰值指标 7 方根幅值 15 裕度指标 8 绝对平均幅值 16 歪度指标 表 2 声发射事件特征
Table 2. Acoustic emission event characteristics
序号 声发射参数 物理意义 1 事件计数 声发射的一次材料局部变化 2 计数 越过门槛信号的振荡次数 3 能量计数 信号检波包络线下的面积 表 3 卷积运算过程
Table 3. Convolution operation process
运算过程 map尺寸 输入层 22×22 卷积层: 卷积核尺寸3×3步长为1 20×20 池化层: 核尺寸2×2步长为2 10×10 卷积层: 卷积核尺寸3×3步长为1 8×8 池化层: 核尺寸2×2步长为2 4×4 表 4 聚类特征指标
Table 4. Convolution operation process
聚类数 PC SC 2 0.955 0 0.008 0 3 0.868 3 0.001 4 4 0.924 8 2.444×10-4 5 0.967 0 1.467×10-5 6 0.853 4 9.503×10-6 7 0.851 0 1.210×10-6 8 0.832 3 2.155×10-5 表 5 结合声发射信号轴承RUL预测结果准确率
Table 5. Accuracy of bearing RUL prediction results
轴承号 80%第1阶段 60%第2阶段 40%第3阶段 20%第4阶段 0%第5阶段 6 83.125 83.75 82.5 85.625 86.875 7 82.5 83.125 81.875 84.375 86.25 8 84.375 85 86.25 85 86.875 表 6 结合加速度信号轴承RUL预测结果准确率
Table 6. Accuracy of bearing RUL prediction results
轴承号 80%第1阶段 60%第2阶段 40%第3阶段 20%第4阶段 0%第5阶段 6 72.125 72.75 72.5 71.625 74.875 7 71.625 72.375 71.875 72.375 74.75 8 72.375 73.125 72.125 72.625 74.625 -
[1] 申中杰, 陈雪峰, 何正嘉, 等. 基于相对特征和多变量支持向量机的滚动轴承剩余寿命预测[J]. 机械工程学报, 2013, 49(2): 183-189. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201302030.htmSHEN Z J, CHEN X F, HE Z J, et al. Remaining life predictions of rolling bearing based on relative features and multivariable support vector machine[J]. Journal of Mechanical Engineering, 2013, 49(2): 183-189. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201302030.htm [2] 文娟, 高宏力. 一种基于UPF的轴承剩余寿命预测方法[J]. 振动与冲击, 2018, 37(24): 208-213. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201824032.htmWEN J, GAO H L. Remaining useful life prediction of bearings with the unscented particle filter approach[J]. Journal of Vibration and Shock, 2018, 37(24): 208-213. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201824032.htm [3] 张星辉, 康建设, 赵劲松, 等. 基于混合高斯输出贝叶斯信念网络模型的设备退化状态识别与剩余使用寿命预测方法研究[J]. 振动与冲击, 2014, 33(8): 171-179. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201408030.htmZHANG X H, KANG J S, ZHAO J S, et al. Equipment degradation state identification and residual life prediction based on MoG-BBN[J]. Journal of Vibration and Shock, 2014, 33(8): 171-179. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201408030.htm [4] KUNDU P, DARPE A K, KULKARNI M S. Weibull accelerated failure time regression model for remaining useful life prediction of bearing working under multiple operating conditions[J]. Mechanical Systems and Signal Processing, 2019, 134: 106302. doi: 10.1016/j.ymssp.2019.106302 [5] REN L, CUI J, SUN Y Q, et al. Multi-bearing remaining useful life collaborative prediction: a deep learning approach[J]. Journal of Manufacturing Systems, 2017, 43: 248-256. doi: 10.1016/j.jmsy.2017.02.013 [6] HINCHI A Z, TKIOUAT M. Rolling element bearing remaining useful life estimation based on a convolutional long-short-term memory network[J]. Procedia Computer Science, 2018, 127: 123-132. doi: 10.1016/j.procs.2018.01.106 [7] AYE S A, HEYNS P S. An integrated Gaussian process regression for prediction of remaining useful life of slow speed bearings based on acoustic emission[J]. Mechanical Systems and Signal Processing, 2017, 84: 485-498. doi: 10.1016/j.ymssp.2016.07.039 [8] WANG Z H, WU X, LIU X Q, et al. Research on feature extraction algorithm of rolling bearing fatigue evolution stage based on acoustic emission[J]. Mechanical Systems and Signal Processing, 2018, 113: 271-284. doi: 10.1016/j.ymssp.2017.08.001 [9] 邱晓梅, 隋文涛, 王峰, 等. 基于相关系数和BP神经网络的轴承剩余寿命预测[J]. 组合机床与自动化加工技术, 2019(4): 63-65. https://www.cnki.com.cn/Article/CJFDTOTAL-ZHJC201904015.htmQIU X M, SUI W T, WANG F, et al. Remaining life prediction of bearing based on correlation coefficient and BP neural network[J]. Modular Machine Tool & Automatic Manufacturing Technique, 2019(4): 63-65. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZHJC201904015.htm [10] 张雨琦, 邹金慧, 马军. 多退化变量灰色预测模型的滚动轴承剩余寿命预测[J]. 探测与控制学报, 2019, 41(3): 112-120. https://www.cnki.com.cn/Article/CJFDTOTAL-XDYX201903021.htmZHANG Y Q, ZOU J H, MA J. Rolling bearing residual life prediction based on grey prediction model with multiple degenerate variables[J]. Journal of Detection & Control, 2019, 41(3): 112-120. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XDYX201903021.htm [11] 徐东, 徐永成, 陈循, 等. 滚动轴承加速寿命试验技术研究[J]. 国防科技大学学报, 2010, 32(6): 122-129. https://www.cnki.com.cn/Article/CJFDTOTAL-GFKJ201006023.htmXU D, XU Y C, CHEN X, et al. Research on accelerated life test for rolling element bearings[J]. Journal of National University of Defense Technology, 2010, 32(6): 122-129. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-GFKJ201006023.htm [12] 王付广, 李伟, 郑近德, 等. 基于多频率尺度模糊熵和ELM的滚动轴承剩余寿命预测[J]. 噪声与振动控制, 2018, 38(1): 188-192. https://www.cnki.com.cn/Article/CJFDTOTAL-ZSZK201801038.htmWANG F G, LI W, ZHENG J D, et al. Prediction of remaining life of rolling bearings based on multi-frequency scale fuzzy entropy and ELM[J]. Noise and Vibration Control, 2018, 38(1): 188-192. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZSZK201801038.htm [13] QIAN Y N, YAN R Q, GAO R X. A multi-time scale approach to remaining useful life prediction in rolling bearing[J]. Mechanical Systems and Signal Processing, 2017, 83: 549-567. [14] ZHAO M H, TANG B P, TAN Q. Bearing remaining useful life estimation based on time-frequency representation and supervised dimensionality reduction[J]. Measurement, 2016, 86: 41-55. [15] 王之海. 基于声发射的球轴承疲劳演化特征提取研究[D]. 昆明: 昆明理工大学, 2017.WANG Z H. Feature extraction of ball bearing fatigue evolution with acoustic emission[D]. Kunming: Kunming University of Science and Technology, 2017. (in Chinese)