A meta transfer learning method for gearbox fault diagnosis with limited data

被引:7
作者
She, Daoming [1 ]
Yang, Zhichao [1 ]
Duan, Yudan [1 ]
Yan, Xiaoan [2 ]
Chen, Jin [1 ]
Li, Yaoming [3 ]
机构
[1] Jiangsu Univ, Sch Mech Engn, Zhenjiang 212013, Peoples R China
[2] Nanjing Forestry Univ, Sch Mechatron Engn, Nanjing 210037, Peoples R China
[3] Jiangsu Univ, Sch Agr Engn, Zhenjiang 212013, Peoples R China
基金
中国国家自然科学基金;
关键词
meta learning; transfer learning; fault diagnosis; gearbox; limited data; NEURAL-NETWORK; MACHINE; ENTROPY;
D O I
10.1088/1361-6501/ad4665
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
Intelligent diagnosis of mechanical faults is an important means to guarantee the safe maintenance of equipment. Cross domain diagnosis may lack sufficient measurement data as support, and this bottleneck is particularly prominent in high-end manufacturing. This paper presents a few-shot fault diagnosis methodology based on meta transfer learning for gearbox. To be specific, firstly, the subtasks for transfer diagnosis are constructed, and then joint distribution adaptation is conducted to align the two domain distributions; secondly, through adaptive manifold regularization, the data of target working condition is further utilized to explore the potential geometric structure of the data distribution. Meta stochastic gradient descent is explored to dynamically adjust the model's parameter based on the obtained task information to obtain better generalization performance, ultimately to achieve transfer diagnosis of gearbox faults with few samples. The effectiveness of the approach is supported by the experimental datasets of the gearbox.
引用
收藏
页数:14
相关论文
共 44 条
[11]   Acoustic spectral imaging and transfer learning for reliable bearing fault diagnosis under variable speed conditions [J].
Hasan, Md Junayed ;
Islam, M. M. Manjurul ;
Kim, Jong-Myon .
MEASUREMENT, 2019, 138 :620-631
[12]   Double-level adversarial domain adaptation network for intelligent fault diagnosis [J].
Jiao, Jinyang ;
Lin, Jing ;
Zhao, Ming ;
Liang, Kaixuan .
KNOWLEDGE-BASED SYSTEMS, 2020, 205
[13]   Knowledge mapping-based adversarial domain adaptation: A novel fault diagnosis method with high generalizability under variable working conditions [J].
Li, Qi ;
Shen, Changqing ;
Chen, Liang ;
Zhu, Zhongkui .
MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2021, 147
[14]   Domain adaptive networks with limited data for rotating machinery fault diagnosis: a case of study of gears [J].
Li, Xueyi ;
Yu, Tianyu ;
He, Qiushi ;
Li, Daiyou ;
Xie, Zhijie ;
Kong, Xiangwei .
MEASUREMENT SCIENCE AND TECHNOLOGY, 2023, 34 (12)
[15]   Intelligent Fault Diagnosis via Semisupervised Generative Adversarial Nets and Wavelet Transform [J].
Liang, Pengfei ;
Deng, Chao ;
Wu, Jun ;
Li, Guoqiang ;
Yang, Zhixin ;
Wang, Yuanhang .
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, 2020, 69 (07) :4659-4671
[16]   Data synthesis using deep feature enhanced generative adversarial networks for rolling bearing imbalanced fault diagnosis [J].
Liu, Shaowei ;
Jiang, Hongkai ;
Wu, Zhenghong ;
Li, Xingqiu .
MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2022, 163
[17]   A conditional variational autoencoding generative adversarial networks with self-modulation for rolling bearing fault diagnosis [J].
Liu, Yunpeng ;
Jiang, Hongkai ;
Wang, Yanfeng ;
Wu, Zhenghong ;
Liu, Shaowei .
MEASUREMENT, 2022, 192
[18]  
Ma B, 2020, International Journal of Information and Communication Technology, V16, P176, DOI [10.1504/ijict.2020.105612, 10.1504/IJICT.2020.105612, DOI 10.1504/IJICT.2020.105612]
[19]   An interpretable data augmentation scheme for machine fault diagnosis based on a sparsity-constrained generative adversarial network [J].
Ma, Liang ;
Ding, Yu ;
Wang, Zili ;
Wang, Chao ;
Ma, Jian ;
Lu, Chen .
EXPERT SYSTEMS WITH APPLICATIONS, 2021, 182
[20]   A hybrid intelligent gearbox fault diagnosis method based on EWCEEMD and whale optimization algorithm-optimized SVM [J].
Men, Zhihui ;
Hu, Chaoqun ;
Li, Yong-Hua ;
Bai, Xiaoning .
INTERNATIONAL JOURNAL OF STRUCTURAL INTEGRITY, 2023, 14 (02) :322-336