-
摘要: 针对现有机车轴承诊断方法存在故障特征提取不理想、诊断精度低等问题，提出了一种基于深度时频特征的机车轴承故障诊断新方法；利用双通道一维和二维卷积神经网络(CNN)分别对输入的一维原始信号和连续小波变换(CWT)提取的二维时频信号进行深度特征提取；为使输入的一维原始信号简单而有效地反映出信号在时域的全局特征，上通道使用一维CNN，为使输入的二维时频域信号能多角度地反映出信号的细微局部变化，下通道使用二维CNN；在融合层中将上下通道特征自动融合成一个新的深度时频特征，并将提取到的深度融合时频特征经归一化指数函数进行故障分类识别；在此基础上，分析了某局机务段实测的7种机车轴承数据，验证了本文方法的实际工程应用价值。研究结果表明：基于深度时频特征的机车轴承故障诊断方法对7种机车轴承故障的平均诊断精度达到了100%，与一维CNN模型、二维CNN模型和支持向量机(SVM)模型相比，平均诊断精度分别提高了0.7%、1.9%和2.2%；本文方法提取的深度时频特征中每类故障分布间隔规则有序，类内间距很小，而单个一维CNN模型和二维CNN模型提取的特征的每类故障分布间隔不规则，类内间距较大，说明基于深度时频特征的机车轴承故障诊断方法提取深度特征的能力优越，是一种解决机车轴承故障诊断问题的有效模型。Abstract: To address the problems such as the unsatisfactory fault feature extraction and low diagnostic accuracy of existing locomotive bearing diagnosis methods, a new method for diagnosing locomotive bearing faults was developed based on the deep time-frequency features. Dual-channel one-dimensional and two-dimensional convolutional neural networks (CNNs) were separately adopted to extract the deep features from the input one-dimensional original and two-dimensional time-frequency signals extracted by the continuous wavelet transform (CWT). A one-dimensional CNN was employed for the upper channel such that the input one-dimensional original signals could effectively reflect the global characteristics of the signals in the time domain. A two-dimensional CNN was applied for the lower channel such that the input two-dimensional time-frequency domain signals could reflect the subtle local changes in the signals from multiple angles. The upper- and lower-channel features were automatically fused in the fusion layer into a new deep time-frequency feature. Then, the extracted deep fusion time-frequency features were classified and identified by a normalized exponential function. Finally, seven types of locomotive bearing data measured in a locomotive depot were analyzed to verify the practical engineering application value of this method. Research results indicate that the average diagnosis accuracies of the proposed method for the seven types of locomotive bearing faults are as high as 100%. Compared with the one-dimensional CNN model, two-dimensional CNN model, and support vector machine (SVM) model, the average diagnosis accuracy of the proposed model increases by 0.7%, 1.9%, and 2.2%, respectively. The distribution intervals of each fault type in the deep time-frequency features are regular and orderly, and the intra-class spacing is very small. Conversely, the features extracted by the single one-dimensional and two-dimensional CNN models exhibit irregular distribution intervals for all fault types, and the intra-class spacing is large. This verifies the superiority of the proposed model in extracting deep features. Therefore, it is an effective model to address the issues in the locomotive bearing fault diagnosis. 4 tabs, 17 figs, 30 refs.
-
图 3 深度时频特征表示的机车轴承故障诊断模型
Figure 3. Locomotive bearing fault diagnosis model presented by deep time-frequency features
图 6 七种机车轴承故障状态时域波形
Figure 6. Time domain waveforms of seven types of locomotive bearing fault states
图 7 七种机车轴承故障状态小波时频
Figure 7. Wavelet time-frequencies of seven types of locomotive bearing fault states
图 8 训练样本与验证样本的损失函数和准确率变化
Figure 8. Changes in loss functions and accuracies of training and verification samples
图 15 二维CNN全连接层可视聚类
Figure 15. Two-dimensional CNN full connection layer visual clustering
图 16 一维CNN全连接层可视聚类
Figure 16. One-dimensional CNN full connection layer visual clustering
表 1 模型结构参数
Table 1. Model structure parameters
编号 上通道层 参数 步长 输出大小 下通道层 参数 步长 输出大小 1 卷积层1 20 5 126×64 卷积层1 4×4 4 8×8×64 2 池化层1 4 4 31×64 池化层1 2×2 2 4×4×64 3 卷积层2 5 2 14×128 卷积层2 2×2 2 2×2×128 4 池化层2 2 1 7×128 池化层2 2×2 1 1×1×128 5 平整层 896 896 平整层 128 128 6 全连接层 128 128 全连接层 128 128 7 融合层 256 8 分类层 7 
下载: 导出CSV
表 2 故障类型与数量
Table 2. Fault types and numbers
编号 故障类型 训练集 测试集 验证集 C1 保持架滚动体复合故障 245 70 35 C2 保持架轻度故障 245 70 35 C3 滚动体轻度故障 245 70 35 C4 正常 245 70 35 C5 外圈中度故障 245 70 35 C6 外圈重度故障 245 70 35 C7 内圈轻度故障 245 70 35
下载: 导出CSV
表 3 模型最终准确率
Table 3. Final accuracies of model
编号 错分
类数准确率/
%编号 错分
类数准确率/
%1 0 100 6 0 100 2 0 100 7 0 100 3 0 100 8 0 100 4 0 100 9 0 100 5 0 100 10 0 100
下载: 导出CSV
表 4 诊断结果
Table 4. Diagnostic results
方法 平均准确率/% 平均误判数 本文方法 100.0 0.0 一维CNN模型 99.3 3.5 二维CNN模型 98.1 10.0 SVM 97.8 11.0
下载: 导出CSV
-
[1] 郑近德, 程军圣, 杨宇. 多尺度排列熵及其在滚动轴承故障诊断中的应用[J]. 中国机械工程, 2013, 24(19): 2641-2646. doi: 10.3969/j.issn.1004-132X.2013.19.017ZHENG Jin-de, CHENG Jun-sheng, YANG Yu. Multi-scale permutation entropy and its applications to rolling bearing fault diagnosis[J]. China Mechanical Engineering, 2013, 24(19): 2641-2646. (in Chinese) doi: 10.3969/j.issn.1004-132X.2013.19.017 [2] 邵海东, 张笑阳, 程军圣, 等. 基于提升深度迁移自动编码器的轴承智能故障诊断[J]. 机械工程学报, 2020, 56(9): 84-90. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB202009011.htmSHAO Hai-dong, ZHANG Xiao-yang, CHENG Jun-sheng, et al. Intelligent fault diagnosis of bearing using enhanced deep transfer auto-encoder[J]. Journal of Mechanical Engineering, 2020, 56(9): 84-90. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB202009011.htm [3] 邓飞跃, 刘鹏飞, 陈恩利, 等. 基于自适应频率窗经验小波变换的列车轮对轴承多故障诊断[J]. 铁道学报, 2019, 41(5): 55-63. doi: 10.3969/j.issn.1001-8360.2019.05.007DENG Fei-yue, LIU Peng-fei, CHEN En-li, et al. Multiple fault diagnosis of train wheelset bearing based on adaptive frequency window empirical wavelet transform[J]. Journal of the China Railway Society, 2019, 41(5): 55-63. (in Chinese) doi: 10.3969/j.issn.1001-8360.2019.05.007 [4] 李永健, 宋浩, 刘吉华, 等. 基于改进多尺度排列熵的列车轴箱轴承诊断方法研究[J]. 铁道学报, 2020, 42(1): 33-39. doi: 10.3969/j.issn.1001-8360.2020.01.005LI Yong-jian, SONG Hao, LIU Ji-hua, et al. A study on fault diagnosis method for train axle box bearing based on modified multiscale permutation entropy[J]. Journal of the China Railway Society, 2020, 42(1): 33-39. (in Chinese) doi: 10.3969/j.issn.1001-8360.2020.01.005 [5] 孟宗, 胡猛, 谷伟明, 等. 基于LMD多尺度熵和概率神经网络的滚动轴承故障诊断方法[J]. 中国机械工程, 2016, 27(4): 433-437. doi: 10.3969/j.issn.1004-132X.2016.04.002MENG Zong, HU Meng, GU Wei-ming, et al. Rolling bearing fault diagnosis method based on LMD multi-scale entropy and probabilistic neural network[J]. China Mechanical Engineering, 2016, 27(4): 433-437. (in Chinese) doi: 10.3969/j.issn.1004-132X.2016.04.002 [6] 梁瑜, 贾利民, 蔡国强, 等. 滚动轴承的非线性动力学故障模型研究[J]. 中国铁道科学, 2014, 35(1): 98-104. doi: 10.3969/j.issn.1001-4632.2014.01.16LIANG Yu, JIA Li-min, CAI Guo-qiang, et al. Research on nonlinear dynamics fault model of rolling bearing[J]. China Railway Science, 2014, 35(1): 98-104. (in Chinese) doi: 10.3969/j.issn.1001-4632.2014.01.16 [7] 朱亚军, 胡建钦, 李武, 等. 基于频域窗函数的短时傅里叶变换及其在机械冲击特征提取中的应用[J]. 机床与液压, 2021, 49(18): 177-182. doi: 10.3969/j.issn.1001-3881.2021.18.035ZHU Ya-jun, HU Jian-qin, LI Wu, et al. Short-time Fourier transform based on frequency-domain window function and its application in mechanical impulse feature extraction[J]. Machine Tool and Hydraulics, 2021, 49(18): 177-182. (in Chinese) doi: 10.3969/j.issn.1001-3881.2021.18.035 [8] SHEN Fei, CHEN Chao, XU Jia-wen, et al. A fast multi-tasking solution: NMF-theoretic co-clustering for gear fault diagnosis under variable working conditions[J]. Chinese Journal of Mechanical Engineering, 2020, 33(1): 16. doi: 10.1186/s10033-020-00437-3 [9] ZHAO Wan-lin, WANG Zi-li, MA Jian, et al. Fault diagnosis of a hydraulic pump based on the CEEMD-STFT time-frequency entropy method and multiclass svm classifier[J]. Shock and Vibration, 2016, 2016: 2609856. [10] BANERJEE T P, DAS S. Multi-sensor data fusion using support vector machine for motor fault detection[J]. Information Sciences, 2012, 217: 96-107. doi: 10.1016/j.ins.2012.06.016 [11] 黄驰城. 结合时频分析和卷积神经网络的滚动轴承故障诊断优化方法研究[D]. 杭州: 浙江大学, 2019.HUANG Chi-cheng. Research on fault diagnosis and optimization method of rolling bearing based on time-frequency analysis and convolutional neural network[D]. Hangzhou: Zhejiang University, 2019. (in Chinese) [12] 乔志城, 刘永强, 廖英英. 改进经验小波变换与最小熵解卷积在铁路轴承故障诊断中的应用[J]. 振动与冲击, 2021, 40(2): 81-90. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202102011.htmQIAO Zhi-cheng, LIU Yong-qiang, LIAO Ying-ying. Application of improved wavelet transform and minimum entropy deconvolution in railway bearing fault diagnosis[J]. Journal of Vibration and Shock, 2021, 40(2): 81-90. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202102011.htm [13] 黄文艺. 基于特征优化与自主学习的滚动轴承故障诊断与性能退化评估[D]. 长沙: 湖南大学, 2019.HUANG Wen-yi. Fault diagnosis and performance degradation assessment of rolling bearing based on feature optimization and adaptive learning[D]. Changsha: Hunan University, 2019. (in Chinese) [14] 张梅军, 唐建, 陈江海. 基于连续小波灰度图的变速箱故障诊断[J]. 振动、测试与诊断, 2007, 27(1): 65-67. doi: 10.3969/j.issn.1004-6801.2007.01.017ZHANG Mei-jun, TANG Jian, CHEN Jiang-hai. Fault diagnosis of gearbox based on continuous wavelet coefficients plot[J]. Journal of Vibration, Measurement and Diagnosis, 2007, 27(1): 65-67. (in Chinese) doi: 10.3969/j.issn.1004-6801.2007.01.017 [15] QU Jin-xiu, ZHANG Zhuo-suo, GONG Teng. A novel intelligent method for mechanical fault diagnosis based on dual-tree complex wavelet packet transform and multiple classifier fusion[J]. Neurocomputing, 2016, 171: 837-853. doi: 10.1016/j.neucom.2015.07.020 [16] 项斌. 基于小波包和支持向量机的机车轴承故障诊断研究[D]. 兰州: 兰州交通大学, 2011.XIANG Bin. Research on locomotive bearing fault diagnosis based on wavelet packet and support vector machines[D]. Lanzhou: Lanzhou Jiaotong University, 2011. (in Chinese) [17] KESKES H, BRAHAM A, LACHIRI Z. Broken rotor bar diagnosis in induction machines through stationary wavelet packet transform and multiclass wavelet SVM[J]. Electric Power Systems Research, 2013, 97: 151-157. doi: 10.1016/j.epsr.2012.12.013 [18] HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504-507. doi: 10.1126/science.1127647 [19] 张西宁, 郭清林, 刘书语. 深度学习技术及其故障诊断应用分析与展望[J]. 西安交通大学学报, 2020, 54(2): 1-13. https://www.cnki.com.cn/Article/CJFDTOTAL-XAJT202012002.htmZHANG Xi-ning, GUO Qing-lin, LIU Shu-yu. Analysis and prospect of deep learning technology and its fault diagnosis application[J]. Journal of Xi'an Jiaotong University, 2020, 54(2): 1-13. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XAJT202012002.htm [20] JANSSENS O, SLAVKOVIKJ V, VERVISCH B, et al. Convolutional neural network based fault detection for rotating machinery[J]. Journal of Sound and Vibration, 2016, 377: 331-345. doi: 10.1016/j.jsv.2016.05.027 [21] INCE T, KIRANYAZ S, EREN L, et al. Real-time motor fault detection by 1-D convolutional neural networks[J]. IEEE Transactions on Industrial Electronics, 2016, 63(11): 7067-7075. doi: 10.1109/TIE.2016.2582729 [22] 马云飞, 贾希胜, 白华军, 等. 基于一维CNN参数优化的压缩振动信号故障诊断[J]. 系统工程与电子技术, 2020, 42(9): 1911-1919. https://www.cnki.com.cn/Article/CJFDTOTAL-XTYD202009006.htmMA Yun-fei, JIA Xi-sheng, BAI Hua-jun, et al. Fault diagnosis of compressed vibration signal based on 1-dimensional CNN with optimized parameters[J]. Systems Engineering and Engineering and Electronics, 2020, 42(9): 1911-1919. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XTYD202009006.htm [23] LEE D, SIU V, CRUZ R, et al. Convolutional neural net and bearing fault analysis[C]//CSREA Press. Proceedings of the International Conference on Data Mining Series. Barcelona: CSREA Press, 2016: 194-200. [24] ZHAO Ming-hang, KANG M S, TANG Bao-ping, et al. Deep residual networks with dynamically weighted wavelet coefficients for fault diagnosis of planetary gearboxes[J]. IEEE Transactions on Industrial Electronics, 2018, 65(5): 4290-4300. doi: 10.1109/TIE.2017.2762639 [25] WANG Jin-jiang, ZHUANG Jun-fei, DUAN Li-xiang, et al. A multi-scale convolution neural network for featureless fault diagnosis[C]//IEEE. Proceedings of 2016 International Symposium on Flexible Automation. New York: IEEE, 2016: 65-70. [26] 王丽华, 谢阳阳, 周子贤, 等. 基于卷积神经网络的异步电机故障诊断[J]. 振动、测试与诊断, 2017, 37(6): 1208-1215, 1283. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCS201706023.htmWANG Li-hua, XIE Yang-yang, ZHOU Zi-xian, et al. Motor fault diagnosis based on convolutional neural network[J]. Journal of Vibration, Measurement and Diagnosis, 2017, 37(6): 1208-1215, 1283. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCS201706023.htm [27] 张龙, 蔡秉桓, 熊国良, 等. 优化指标一致的滚动轴承故障复合诊断方法[J]. 振动与冲击, 2021, 40(9): 237-245. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202109032.htmZHANG Long, CAI Bing-huan, XIONG Guo-liang, et al. Composite fault diagnosis method of rolling bearing based on consistent optimization index[J]. Journal of Vibration and Shock, 2021, 40(9): 237-245. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202109032.htm [28] 李涛, 段礼祥, 张东宁, 等. 自适应卷积神经网络在旋转机械故障诊断中的应用[J]. 振动与冲击, 2020, 39(16): 275-282, 288. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202016037.htmLI Tao, DUAN Li-xiang, ZHANG Dong-ning, et al. Application of adaptive convolutional neural network in rotating machinery fault diagnosis[J]. Journal of Vibration and Shock, 2020, 39(16): 275-282, 288. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ202016037.htm [29] 张淑清, 包红燕, 李盼, 等. 基于RQA与GG聚类的滚动轴承故障识别[J]. 中国机械工程, 2015, 26(10): 1385-1390. doi: 10.3969/j.issn.1004-132X.2015.10.019ZHANG Shu-qing, BAO Hong-yan, LI Pan, et al. Fault diagnosis of rolling bearings based on RQA and GG clustering[J]. China Mechanical Engineering, 2015, 26(10): 1385-1390. (in Chinese) doi: 10.3969/j.issn.1004-132X.2015.10.019 [30] 张立国, 李盼, 李梅梅, 等. 基于ITD模糊熵和GG聚类的滚动轴承故障诊断[J]. 仪器仪表学报, 2014, 35(11): 2624-2632. https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201411027.htmZHANG Li-guo, LI Pan, LI Mei-mei, et al. Fault diagnosis of rolling bearing based on ITD fuzzy entropy and GG clustering[J]. Chinese Journal of Scientific Instrument, 2014, 35(11): 2624-2632. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201411027.htm -
点击查看大图
图(17) / 表(4)
计量
- 文章访问数: 463
- HTML全文浏览量: 233
- PDF下载量: 71
- 被引次数: 0
