A compound fault identification method for gearbox based on Triplet loss
ZHAO Xiaoping1，3, WANG Yifei2, ZHANG Yonghong2, WU Jiaxin1, WANG Lihua2
1.School of Computer and Software, Nanjing University of Information Engineering, Nanjing 210044, China;
2.School of Automation, Nanjing University of Information Engineering, Nanjing 210044, China;
3.Jiangsu Provincial Network Monitoring Center, Nanjing University of Information Engineering, Nanjing 210044, China
Abstract： With the increase in the number of device detection points and the sampling frequency, mechanical health monitoring has entered the era of big data. With its powerful adaptive feature extraction and classification capabilities, deep learning has achieved good results in mechanical data processing. However, most research objects in the field of fault diagnosis are single faults. Multi-fault is difficult to diagnose accurately because of the coupling between various kinds of fault signals and the great influence of changing working conditions on the signals. in the case of a wide variety of composite faults, the traditional softmax classifier cannot complete fault diagnosis accurately and efficiently. To solve this problem, this paper proposes deep metric learning based on triplet loss to complete the diagnostic classification of bearing and gear of gearbox simultaneously. The advantage of this model is to extract the features of fault signals, and then measure the distances between different kinds of faults by Triplet loss, which makes the distances between similar fault features very close, and between different fault features very far, thus completing the diagnosis task efficiently. The experimental results show that the method can accurately diagnose the different faults of the inner bearing and the gear under various working conditions and a large number of samples.
赵晓平1,3，王逸飞 2，张永宏 2，吴家新 1，王丽华 2. 一种基于Triplet loss的齿轮箱复合故障识别方法[J]. 振动与冲击, 2021, 40(5): 46-54.
ZHAO Xiaoping1，3, WANG Yifei2, ZHANG Yonghong2, WU Jiaxin1, WANG Lihua2. A compound fault identification method for gearbox based on Triplet loss. JOURNAL OF VIBRATION AND SHOCK, 2021, 40(5): 46-54.
[1] Yang Z, Yu Z, Xie C, et al. Application of Hilbert–Huang Transform to acoustic emission signal for burn feature extraction in surface grinding process[J]. Measurement, 2014, 47(1): 14-21.
[2] 陈果. 滚动轴承早期故障的特征提取与智能诊断[J]. 航空学报, 2009, 30(2):362-367.
CHEN Guo, Feature Extraction and Intelligent Diagnosis of Early Failure of Rolling Bearings [J]. Journal of Aviation, 2009, 30(2): 362-367.
[3] 徐敏强, 黄文虎, 张嘉钟. 旋转机械高速启动过程振动信号分析方法的研究[J]. 振动工程学报, 2000, 13(2): 216-221.
XU Min-qiang, HUANG Wen-hu, ZHANG Jia-zhong. Research on Vibration Signal Analysis Method of Rotating Machinery in High Speed Start Process [J]. Journal of Vibration Engineering, 2000, 13(2): 216-221.
[4] 段艳杰, 吕宜生, 张杰,等. 深度学习在控制领域的研究现状与展望[J]. 自动化学报, 2016, 42(5):643-654.
DUAN Yan-jie, LV Yi-sheng, ZHANG Jie, et al. Research status and prospects of deep learning in the field of control [J]. Journal of Automation, 2016, 42(5): 643-654.
[5] Chen D , Cao X , Wen F , et al. Blessing of Dimensionality: High-Dimensional Feature and Its Efficient Compression for Face Verification[C]// IEEE Conference on Computer Vision & Pattern Recognition. IEEE, 2013.
[6] Mbo'O C P , Hameyer K . Bearing damage diagnosis by means of the linear discriminant analysis of stator current feature[C]// IEEE International Symposium on Diagnostics for Electrical Machines. IEEE, 2015.
[7] WANG Xing-qing, LI Yan-feng, TING Rui, et al. Bearing fault diagnosis method based on Hilbert envelope spectrum and deep belief network[J]. Journal of Vibroengineering, 2015, 17(3): 1295-1308.
[8] 张芳芳, 贺娟, 李明军. 基于导数优化的BP学习算法的研究综述[J]. 计算机应用研究, 2009, 026(003):809-813.
ZHANG Fang-fang, HE Juan, LI Ming-jun, Overviews on research of back-propagation algorithm based on derivative optimization[J]. Research on computer application, 2009, 026(003):809-813.
[9] Li C, Zurita G, Cerrada M, et al. Multimodal deep support vector classification with homologous features and its application to gearbox fault diagnosis[J]. Neurocomputing, 2015, 168(C):119-127.
[10] Shao H, Jiang H, Zhang X, et al. Rolling bearing fault diagnosis using an optimization deep belief network[J]. Measurement Science & Technology, 2015, 26（11）: 115002
[11] Bhadane M, Ramachandran K I. Bearing fault identification and classification with convolutional neural network[C]// International Conference on Circuit, power and Computing Technologies. IEEE, 2017:1-5.
[12] Jeong H, Park S, Woo S, et al. Rotating Machinery Diagnostics Using Deep Learning on Orbit Plot Images [J]. Procedia Manufacturing, 2016, 5:1107-1118.
[13] ZHANG Wei, PENG Gao-liang, LI Chuan-hao. Advances in Intelligent Information Hiding and Multimedia Signal Processing[M], Taiwan, Springer, 2016:77-84
[14] Huang H B, Huang X R, Li R X, Et Al. Sound quality prediction of vehicle interior noise using deep belief networks[J]. Applied Acoustics, 2016, 113: 149-161.
[15] Tamilselvan P, Wang P. Failure diagnosis using deep belief learning based health state classification[J]. Reliability Engineering & System Safety, 2013, 115（7）: 124-135.
[16] 陈淑梅,余建波.卷积神经网络多变量过程特征学习与故障诊断[J/OL].哈尔滨工业大学学报:1-11[2020-03-23].
[17] Schroff Florian, Kalenichenko D, Philbin J. FaceNet: A unified embedding for face recognition and clustering[J]. 2015:815-823.
[18] 王梦阳, 王华庆, 董方. 基于EVMD-LNMF的复合故障信号分离方法[J]. 振动与冲击, 2019(16).
Wang Meng-yang, Wang Qin-hua, Dong fang. Compound Fault Signal Separation Method Based on EVMD-LNMF[J]. Vibration and shock, 2019(16).
[19] LIU Hong-mei, LI Lian-feng, MA Jian. Rolling Bearing Fault Diagnosis Based on STFT-Deep Learning and Sound Signals[J]. Shock and Vibration, 2016:1-12.
[20] WANG Li-hua, ZHAO Xiao-ping, WU Jia-xin, et al. Motor Fault Diagnosis Based on Short-time Fourier Transform and Convolutional Neural Network[J]. Chinese Journal of Mechanical Engineering, 2017, 30(6): 1357-1368.
[21] ZHAO Xiao-ping, WU Jia-xin, ZHANG Yong-hong, SHI Yun-qing, WANG Li-hua. Fault diagnosis of motor in frequency domain signal by stacked de-noising auto-encoder. Computers, Materials & Continua, 2018, vol. 57, no. 2, pp. 223-242.
