稀疏性多源完全领域适配迁移诊断方法研究
Sparse multi-source entire domain adaptation transfer diagnosis method
- 铁道科学与工程学报 2023年20卷第11期 页码：4438-4450
- 作者机构：
1.西南交通大学 牵引动力国家重点实验室，四川 成都 610031
- 作者简介：
易彩(1987—)，女，四川成都人，助理研究员，博士，从事高速列车轴箱轴承振动特性、故障诊断方法研究；E-mail：yicai@swjtu.edu.cn
- 基金信息：国家自然科学基金资助项目(52272355;51905453);四川省自然科学基金资助项目(2022NSFSC1918);牵引动力国家重点实验室研究基金资助项目(2020TPL-T14);中国博士后科学基金资助项目(2019M663899XB)
DOI：10.19713/j.cnki.43-1423/u.T20222445
中图分类号：
扫 描 看 全 文
邢展,易彩,林建辉.稀疏性多源完全领域适配迁移诊断方法研究[J].铁道科学与工程学报,2023,20(11):4438-4450.
XING Zhan,YI Cai,LIN Jianhui.Sparse multi-source entire domain adaptation transfer diagnosis method[J].Journal of Railway Science and Engineering,2023,20(11):4438-4450.
邢展,易彩,林建辉.稀疏性多源完全领域适配迁移诊断方法研究[J].铁道科学与工程学报,2023,20(11):4438-4450. DOI： 10.19713/j.cnki.43-1423/u.T20222445.
XING Zhan,YI Cai,LIN Jianhui.Sparse multi-source entire domain adaptation transfer diagnosis method[J].Journal of Railway Science and Engineering,2023,20(11):4438-4450. DOI： 10.19713/j.cnki.43-1423/u.T20222445.
摘要
高速列车在行驶过程中，其滚动轴承的工况往往会发生变化，如转向架的轴箱轴承转速变化等，在此情况下，传统的基于非平稳信号分析的滚动轴承故障诊断技术很难对故障进行有效检测。为了在变化的工况下实现故障类型的自适应识别，基于深度迁移学习的滚动轴承智能故障诊断技术逐渐被应用于轴承故障诊断领域中。然而，深度迁移学习智能故障诊断技术在工程应用中仍存在模型复杂度高的问题。此外，传统领域适配方法在进行迁移诊断时主要进行整体领域差异的对齐，忽略了不同域下同一故障状态的子域特征分布，导致模型的泛化能力差。同时，领域适配在滚动轴承智能故障诊断中的应用主要集中在单源领域，当源领域和目标领域数据分布差异过大时，单个源领域学习到的数据特征有限，可能无法达到较高的可迁移性。为改善上述问题，提出一种稀疏性多源完全领域适配迁移诊断方法，为验证所提算法的有效性，采用不同的故障数据建立不同的故障诊断场景并进行分析。结果表明，该方法中提出的周期循环稀疏设计模式使得卷积层和全连接层的权重矩阵包含大量规则排列的零权重参数，能够有效降低模型的复杂度。同时，该方法在进行数据特征的迁移时，考虑了全局和局部差异的对齐，能够一定程度改善模型的泛化能力。此外，通过不同数目源领域的试验验证分析，表明多源领域适配方法具有较好的迁移诊断效果，对于迁移诊断技术的实际应用具有一定的指导意义。
Abstract
In the running process of high-speed train, the working condition of the rolling bearing is usually changeable, such as the rotating speed change of the axle-box bearing of the bogie. In this case, the traditional rolling bearing fault diagnosis technology based on non-stationary signal analysis is difficult to detect the fault effectively. In order to realize the adaptive identification of fault types under the changeable working conditions, rolling bearing fault diagnosis technology based on deep transfer learning is gradually applied in the field of bearing fault diagnosis. However, deep transfer learning intelligent fault diagnosis technology still has the problem of high model complexity in engineering application. In addition, the traditional domain adaptation methods mainly align the global discrepancy in transfer diagnosis, ignoring the subdomain feature distribution of the same fault status in different domains and resulting in poor generalization ability of the model. Meanwhile, the application of domain adaptation in the rolling bearing intelligent fault diagnosis mainly focuses on the single source domain adaptation. When the data distribution discrepancy between the source and target domain is large, the data features learned by the single source domain are limited, which may not achieve high transferability. In order to solve the above problems, a sparse multi-source entire domain adaptation transfer diagnosis method was proposed. In order to verify the effectiveness of the proposed method, different transfer scenarios were established and analyzed with different fault datasets. The results show that the proposed periodic cyclic sparse design mode in this method can make the weight matrix of the convolutional layer and the fully-connected layer contain a large number of regularly arranged zero-weight parameters, which effectively reduces the complexity of the model. Meanwhile, this method considers the alignment of global and local discrepancy when transferring data features, which can improve the generalization ability of the model to some extent. In addition, the experimental verification analysis of different number of source domains shows that the multi-source domain adaptation method has better transfer diagnosis effect, which has certain guiding significance for the practical application of transfer diagnosis technology.
关键词
高速列车滚动轴承周期循环稀疏多源领域适配
Keywords
high speed trainrolling bearingperiodic cyclic sparsemulti-source domain adaptation
references
张龙, 闫乐玮, 熊国良, 等. 基于EWT-MCKD的机车轮对轴承故障诊断[J]. 铁道科学与工程学报, 2021, 18(10): 2722-2732.
ZHANG Long, YAN Lewei, XIONG Guoliang, et al. Fault diagnosis of locomotive wheel set bearing based on EWT-MCKD[J]. Journal of Railway Science and Engineering, 2021, 18(10): 2722-2732.
KHAN S, YAIRI T. A review on the application of deep learning in system health management[J]. Mechanical Systems and Signal Processing, 2018, 107: 241-265.
蒋雨良, 曾大懿, 邹益胜, 等. 基于多任务学习的轨道车辆轴承异常检测方法[J]. 铁道科学与工程学报, 2021, 18(5): 1267-1276.
JIANG Yuliang, ZENG Dayi, ZOU Yisheng, et al. Abnormal detection method of rail vehicle bearing based on multi task learning[J]. Journal of Railway Science and Engineering, 2021, 18(5): 1267-1276.
LIU Ruonan, YANG Boyuan, ZIO E, et al. Artificial intelligence for fault diagnosis of rotating machinery: a review[J]. Mechanical Systems and Signal Processing, 2018, 108: 33-47.
于洋, 马军, 王晓东, 等. 基于GST与改进CNN的滚动轴承智能故障诊断[J]. 铁道科学与工程学报, 2022, 19(7): 2050-2060.
YU Yang, MA Jun, WANG Xiaodong, et al. Intelligent fault diagnosis of rolling bearings based on GST and improved CNN[J]. Journal of Railway Science and Engineering, 2022, 19(7): 2050-2060.
ZHAO Zhibin, ZHANG Qiyang, YU Xiaolei, et al. Applications of unsupervised deep transfer learning to intelligent fault diagnosis: a survey and comparative study[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 1-28.
ZHAO Bo, ZHANG Xianmin, ZHAN Zhenhui, et al. Deep multi-scale convolutional transfer learning network: a novel method for intelligent fault diagnosis of rolling bearings under variable working conditions and domains[J]. Neurocomputing, 2020, 407: 24-38.
WANG Yu, SUN Xiaojie, LI Jie, et al. Intelligent fault diagnosis with deep adversarial domain adaptation[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 1-9.
YANG Bin, LEI Yaguo, JIA Feng, et al. An intelligent fault diagnosis approach based on transfer learning from laboratory bearings to locomotive bearings[J]. Mechanical Systems and Signal Processing, 2019, 122: 692-706.
HAN Te, LIU Chao, YANG Wenguang, et al. Deep transfer network with joint distribution adaptation: a new intelligent fault diagnosis framework for industry application[J]. ISA Transactions, 2020, 97: 269-281.
DETTMERS T, ZETTLEMOYER L. Sparse networks from scratch: faster training without losing performance[EB/OL].(2019-06-10)[2022-12-28]. https://doi.org/10.48550/arXiv.1907.04840https://doi.org/10.48550/arXiv.1907.04840.
SCARDAPANE S, COMMINIELLO D, HUSSAIN A, et al. Group sparse regularization for deep neural networks[J]. Neurocomputing, 2017, 241: 81-89.
XU Ruijia, CHEN Ziliang, ZUO Wangmeng, et al. Deep cocktail network: multi-source unsupervised domain adaptation with category shift[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City, UT, USA. IEEE, 2018: 3964-3973.
ZHAO Han, ZHANG Shanghang, WU Guanhang, et al. Multiple source domain adaptation with adversarial training of neural networks[EB/OL]. (2017-05-26)[2022-12-28]. https://arxiv.org/abs/1705.09684https://arxiv.org/abs/1705.09684.
KUNDU S, NAZEMI M, PEDRAM M, et al. Pre-defined sparsity for low-complexity convolutional neural networks[J]. IEEE Transactions on Computers, 2020, 69(7): 1045-1058.
GRETTON A, BORGWARDT K M, RASCH M, et al. A kernel method for the two-sample-problem[C]// Proceedings of the 19th International Conference on Neural Information Processing Systems. December 4 - 7, 2006, Canada. New York: ACM, 2006: 513-520.
GRETTON A, SRIPERUMBUDUR B, SEJDINOVIC D, et al. Optimal kernel choice for large-scale two-sample tests[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems. December 3 - 6, 2012, Lake Tahoe, Nevada. New York: ACM, 2012: 1205-1213.
ZHU Yongchun, ZHUANG Fuzhen, WANG Jindong, et al. Deep subdomain adaptation network for image classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(4): 1713-1722.
ZHU Yongchun, ZHUANG Fuzhen, WANG Deqing. Aligning domain-specific distribution and classifier for cross-domain classification from multiple sources[C]//Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence and Thirty-First Innovative Applications of Artificial Intelligence Conference and Ninth AAAI Symposium on Educational Advances in Artificial Intelligence. New York: ACM, 2019: 5989-5996.
ZHANG Zhendong, WANG Xinran, JUNG C. DCSR: dilated convolutions for single image super-resolution[J]. IEEE Transactions on Image Processing: a Publication of the IEEE Signal Processing Society, 2019, 28(4): 1625-1635.
HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 27-30, 2016, Las Vegas, NV, USA. IEEE, 2016: 770-778.
LESSMEIER C, KIMOTHO J, ZIMMER D, et al. Condition monitoring of bearing damage in electromechanical drive systems by using motor current signals of electric motors: a benchmark data set for data-driven classification[C]// European Conference of the Prognostics and Health Management Society. July 6-8, 2016, Bilbao, Spain. 2016: 1-17.
PHMSOCIETY. PHM09 Data Challenge[EB/OL]. (2009-04-10)[2022-12-28]. https://phmsociety.org/data-analysis-competitionhttps://phmsociety.org/data-analysis-competition.
LONG Mingsheng, ZHU Han, WANG Jianmin, et al. Deep transfer learning with joint adaptation networks[C]// Proceedings of the 34th International Conference on Machine Learning-Volume 70. August 6-11, 2017, Sydney, NSW, Australia. New York: ACM, 2017: 2208-2217.
0
浏览量
2
下载量
0
CSCD
- 网络PDF下载
- XML下载
- Meta-XML下载
- BibTeX下载
- Endnote下载
- RefMan 下载
- NoteExpress下载
- NoteFirst下载
关联资源
相关文章
相关作者
相关机构
- 地址：长沙市韶山南路22号 邮编：410075
- 联系电话：0731-82656536 Email：jrse@csu.edu.cn
- 技术支持由北京北大方正电子有限公司提供 湘ICP备09001153号-13
京公网安备11010802024621 - 本系统建议在Chrome、 IE9+ 以上版本浏览器阅读本站内容，360浏览器请切换至极速模式
- Cookies帮助我们提供服务并提供个性化体验。使用本网站，即表示您同意我们使用Cookies