Health assessment method of traction motor bearing based on transfer learning and convolutional neural network
-
摘要: 针对牵引电机轴承健康评估中带标签的全寿命周期振动数据获取与可反映轴承性能退化趋势的健康指标构建困难的问题，提出了一种基于迁移学习和卷积神经网络的牵引电机轴承健康评估方法；采用迁移学习，以带标签的轴承全寿命周期数据集为源域数据，以综合试验台数据为目标域数据，构建数据集；采用欠采样与合成少数类过采样技术对全寿命周期数据集进行扩充与平衡，得到了卷积神经网络训练所需的有效样本数量；在时域和频域上提取描述轴承退化过程的特征，利用卷积神经网络，遵循轴承性能退化规律的浴缸曲线，对基本特征进行融合, 构造了健康评估指标。分析结果表明: 在电机轴承轴电流损伤的健康评估中，所提出的基于迁移学习和卷积神经网络的健康评估方法的准确率为98.17%，遵循直线型、二次函数型和抛物线型退化规律构建健康指标的方法的准确率分别为86.61%、89.56%、91.30%，因此，所提评估方法准确率最大，具有更佳的评估效果，并且实现专家知识与神经网络学习知识的结合，降低了故障特征维度，解决了健康指标构建困难的问题，通过跨设备迁移学习实现了牵引电机轴承的健康评估。Abstract: To address the difficulties in acquiring labeled life-cycle vibration data and constructing health indicators that can reflect the bearing performance degradation trends in the health assessment of traction motor bearing, a health assessment method based on a transfer learning and convolutional neural network model was proposed for implementing the health assessment of traction motor bearing. The labeled bearing life-cycle data set was used as the source domain data by using the transfer learning, and the comprehensive test bench data were used as the target domain data to construct the data set. Undersampling and synthetic minority over-sampling techniques were used to expand and balance the life-cycle data set, and the abundant sample for convolutional neural networks training was obtained. The features describing the bearing degradation process were extracted in time domain and frequency domain. By using convolutional neural network and following the bathtub curve of bearing performance degradation, the health indicator was constructed by fusing the basic characteristics. Analysis results show that in the health assessment of traction motor bearing shaft current damage, the accuracy of the proposed health assessment method based on transfer learning and convolutional neural network is 98.17%. The accuracies of the methods constructing the health indicator according to linear, quadratic function and parabolic degradation law are 86.61%, 89.56% and 91.30%, respectively. Therefore, the proposed health assessment method has higher accuracy and better evaluation effect in the application. Moreover, the combination of expert knowledge and neural network learning knowledge reduces the fault characteristic dimension, solves the difficulty of health indicator construction, and realizes the health assessment of traction motor bearings through cross-device transfer learning.
-
图 2 基于迁移CNN的牵引电机轴承健康评估流程
Figure 2. Health assessment flow of traction motor bearing based on transfer CNN
表 1 XJTU-SY全寿命周期轴承数据集(部分)
Table 1. XJTU-SY full life cycle bearing data set (partial)
工况1 工况2 工况3 XJ11、XJ12、XJ13、XJ15 XJ21、XJ22、XJ23、XJ24、XJ25 XJ31、XJ33 
下载: 导出CSV
表 2 CNN架构
Table 2. CNN architecture
层 类型 参数 训练参数 1 输入层 12×32×1 填充=‘same’，初始学习率为0.001，最大训练次数为80，小批量数为64 2 卷积层1 C=11, N=32, S=1 3 批量归一化层1 N=32 4 线性整流函数激活层1 N=32 5 平均池化层1 C=2, N=32, S=2 6 卷积层2 C=11, N=32, S=2 7 批量归一化层2 N=32 8 线性整流函数激活层2 N=32 9 平均池化层2 C=2, N=32, S=2 10 卷积层3 C=11, N=32, S=2 11 批量归一化层3 N=32 12 线性整流函数激活层3 N=32 13 平均池化层3 C=2, N=32, S=2 14 全连接层 1 15 回归层
下载: 导出CSV
表 3 健康评估数据集
Table 3. Health assessment data sets
源域数据 工况1 XJ111、XJ121、XJ131、XJ151、XJ152、XJ153 工况2 XJ211、XJ212、XJ213、XJ221、XJ222、XJ231、XJ241、XJ251 工况3 XJ311、XJ312、XJ331、XJ332、XJ333 目标域数据 训练集 QY111、QY112、QY113、QY211、QY212、QY213、QY311、QY312、QY313 测试集 QY12、QY22、QY32
下载: 导出CSV
表 4 评估结果
Table 4. Evaluation results
线型 准确率/% RMSE MAE 反双曲正切 98.17 0.039 8 0.031 4 直线型 86.61 0.070 3 0.053 8 二次函数型 89.56 0.063 8 0.049 8 抛物线型 91.30 0.056 8 0.044 0
下载: 导出CSV
-
[1] 张济民, 苏辉, 任乔, 等. 轨道交通永磁同步牵引系统发展概况与关键技术综述[J]. 交通运输工程学报, 2021, 21(6): 63-77. doi: 10.19818/j.cnki.1671-1637.2021.06.005ZHANG Ji-min, SU Hui, REN Qiao, et al. Review on development and key technologies of permanent magnet synchronous traction system for rail transit[J]. Journal of Traffic and Transportation Engineering, 2021, 21(6): 63-77. (in Chinese) doi: 10.19818/j.cnki.1671-1637.2021.06.005 [2] 邓浩衡, 易锦阳, 王烟平. 长沙轨道交通1号线牵引电机轴承电腐蚀原因分析及整改[J]. 电力机车与城轨车辆, 2019, 42(4): 65-67. https://www.cnki.com.cn/Article/CJFDTOTAL-DJJI201904019.htmDENG Hao-heng, YI Jin-yang, WANG Yan-ping. Cause analysis and rectification of electrical corrosion of traction motor for Changsha Rail Transit Line 1[J]. Electric Locomotives and Mass Transit Vehicles, 2019, 42(4): 65-67. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-DJJI201904019.htm [3] 范彪. 广州地铁二号线A5型车牵引电机轴承电腐蚀问题调查分析及解决措施[J]. 机电工程技术, 2017, 46(2): 134-138. https://www.cnki.com.cn/Article/CJFDTOTAL-JXKF201702033.htmFAN Biao. Investigation and analysis of the electric corrosion of the traction motor bearing of the A5 type car in Guangzhou Metro Line 2[J]. Mechanical and Electrical Engineering Technology, 2017, 46(2): 134-138. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXKF201702033.htm [4] 贾正阳, 宋俊杰, 汪久根, 等. 考虑绝缘性能的变频调速电机轴承优化设计[J]. 计算机集成制造系统, 2022, 28(9): 2726-2738. doi: 10.13196/j.cims.2022.09.007JIA Zheng-yang, SONG Jun-jie, WANG Jiu-gen, et al. Optimal design of variable frequency speed regulation motor bearing considering insulation performance[J]. Computer Integrated Manufacturing Systems, 2022, 28(9): 2726-2738. (in Chinese) doi: 10.13196/j.cims.2022.09.007 [5] SHAHRIAR M R, BORGHESANI P, TAN A C C. Electrical signature analysis-based detection of external bearing faults in electromechanical drivetrains[J]. IEEE Transactions on Industrial Electronics, 2018, 65(7): 5941-5950. doi: 10.1109/TIE.2017.2782240 [6] 崔石玉, 朱志宇. 基于参数迁移和一维卷积神经网络的海水泵故障诊断[J]. 振动与冲击, 2021, 40(24): 180-189. doi: 10.13465/j.cnki.jvs.2021.24.022CUI Shi-yu, ZHU Zhi-yu. Seawater pump fault diagnosis based on parameter transfer and one-demensional convolutional neural network[J]. Journal of Vibration and Shock, 2021, 40(24): 180-189. (in Chinese) doi: 10.13465/j.cnki.jvs.2021.24.022 [7] 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. https://www.cnki.com.cn/Article/CJFDTOTAL-XAJT202208001.htmJIA Feng, LI Shi-hao, SHEN Jian-jun, et al. Fault diagnosis of rolling bearings using deep transfer learning and adaptive weighting[J]. Journal of Xi'an Jiaotong University, 2022, 56(8): 1-10. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XAJT202208001.htm [8] RUAN Di-wang, ZHANG Fei-fan, YAN Jian-ping. Transfer learning between different working conditions on bearing fault diagnosis based on data augmentation[J]. IFAC-PapersOnLine, 2021, 54(1): 1193-1199. doi: 10.1016/j.ifacol.2021.08.141 [9] ZHANG Xu, CHEN Zhi-kui, GAO Jing, et al. A two-stage deep transfer learning model and its application for medical image processing in traditional Chinese medicine[J]. Knowledge-Based Systems, 2022, 239: 108060. [10] LI Jiang-kuan, LIN Meng, LI Yan-kai, et al. Transfer learning with limited labeled data for fault diagnosis in nuclear power plants[J]. Nuclear Engineering and Design, 2022, 390: 111690. [11] WANG Dong, TSUI K L, MIAO Qiang. Prognostics and health management: a review of vibration based bearing and gear health indicators[J]. IEEE Access, 2018, 6: 665-676. [12] LEI Ya-guo, LI Nai-peng, GUO Liang, et al. Machinery health prognostics: a systematic review from data acquisition to RUL prediction[J]. Mechanical Systems and Signal Processing, 2018, 104: 799-834. [13] MINHAS A S, SINGH S. A new bearing fault diagnosis approach combining sensitive statistical features with improved multiscale permutation entropy method[J]. Knowledge-Based Systems, 2021, 218: 106883. [14] 陈骏杰, 师蔚, 胡定玉. 基于IMF聚合与SVD的城轨车辆牵引电机轴承故障诊断[J]. 测控技术, 2017, 36(1): 14-17, 22. https://www.cnki.com.cn/Article/CJFDTOTAL-IKJS201701004.htmCHEN Jun-jie, SHI Wei, HU Ding-yu. Fault diagnosis for traction motor rolling bearings in urban rail vehicles based on IMF aggregation and SVD[J]. Measurement and Control Technology, 2017, 36(1): 14-17, 22. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-IKJS201701004.htm [15] RAI A, UPADHYAY S H. Bearing performance degradation assessment based on a combination of empirical mode decomposition and K-medoids clustering[J]. Mechanical Systems and Signal Processing, 2017, 93: 16-29. [16] XU F, SONG X B, TSUI K L, et al. Bearing performance degradation assessment based on ensemble empirical mode decomposition and affinity propagation clustering[J]. IEEE Access, 2019, 7: 54623-54637. [17] LI Yao-long, LI Hong-ru, WANG Bing, et al. Rolling element bearing performance degradation assessment using variational mode decomposition and Gath-Geva clustering time series segmentation[J]. International Journal of Rotating Machinery, 2017, 2017: 2598169. [18] FENG Fu-zhou, WU Chun-zhi, ZHU Jun-zhen, et al. Research on multitask fault diagnosis and weight visualization of rotating machinery based on convolutional neural network[J]. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2020, 42: 603. [19] XIE Y, ZHANG T. Fault diagnosis for rotating machinery based on convolutional neural network and empirical mode decomposition[J]. Shock and Vibration, 2017, 2017: 3084197. [20] 刘世林, 陈里里. 基于VMD-SPWVD-CNN的滚动轴承故障智能诊断[J]. 组合机床与自动化加工技术, 2022, 2022(4): 62-65, 69. https://www.cnki.com.cn/Article/CJFDTOTAL-ZHJC202204014.htmLIU Shi-lin, CHEN Li-li. Intelligent fault diagnosis of rolling bearing based on VMD-SPWVD-CNN[J]. Modular Machine Tool and Automatic Manufacturing Technique, 2022, 2022(4): 62-65, 69. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZHJC202204014.htm [21] 宋庭新, 黄继承, 刘尚奇, 等. 小样本下基于DWT和2D-CNN的齿轮故障诊断方法[J]. 计算机集成制造系统, https://kns.cnki.net/kcms/detail/11.5946.TP.20230423.1131.004.html .SONG Ting-xin, HUANG Ji-cheng, LIU Shang-qi, et al. Gear fault diagnosis method based on DWT and 2D-CNN in small samples[J]. Computer Integrated Manufacturing Systems,https://kns.cnki.net/kcms/detail/11.5946.TP.20230423.1131.004.html . (in Chinese)[22] CHAWLA N V, BOWYER K W, HALL L O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16: 321-357. [23] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324. [24] 庄福振, 罗平, 何清, 等. 迁移学习研究进展[J]. 软件学报, 2015, 26(1): 26-39. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201501003.htmZHUANG Fu-zhen, LUO Ping, HE Qing, et al. Survey on transfer learning research[J]. Journal of Software, 2015, 26(1): 26-39. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201501003.htm [25] 王凯, 李元辉. 迁移学习在机械设备预测性维护领域的应用综述[J]. 中国仪器仪表, 2019(12): 64-68. https://www.cnki.com.cn/Article/CJFDTOTAL-ZYQB201912016.htmWANG Kai, LI Yuan-hui. Summary of application of transfer learning in predictive maintenance of machinery and equipment[J]. China Instrumentation, 2019(12): 64-68. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZYQB201912016.htm [26] QIAN Chen-hui, ZHU Jun-jun, SHEN Ye-hu, et al. Deep transfer learning in mechanical intelligent fault diagnosis: application and challenge[J]. Neural Processing Letters, 2022, 54(3): 2509-2531. [27] ZHUANG Fu-zhen, QI Zhi-yuan, DUAN Ke-yu, et al. A comprehensive survey on transfer learning[J]. Proceedings of the IEEE, 2021, 109(1): 43-76. [28] 雷亚国, 韩天宇, 王彪, 等. XJTU-SY滚动轴承加速寿命试验数据集解读[J]. 机械工程学报, 2019, 55(16): 1-6. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201916001.htmLEI Ya-guo, HAN Tian-yu, WANG Biao, et al. XJTU-SY rolling element bearing accelerated life test datasets: a tutoria[J]. Journal of Mechanical Engineering, 2019, 55(16): 1-6. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201916001.htm -
点击查看大图
图(9) / 表(4)
计量
- 文章访问数: 273
- HTML全文浏览量: 124
- PDF下载量: 77
- 被引次数: 0
