Cross-domain Fault Diagnosis of Gearboxes Based on Reinforcement Class-level Matching Network

RAO Lang; DENG Aidong; DENG Minqiang; WU Yifan; XIA Weiping; HU Qinyi

doi:10.19805/j.cnki.jcspe.2026.240685

English Version Journal Login

您当前的位置：

首页 >

文章列表页 >

Cross-domain Fault Diagnosis of Gearboxes Based on Reinforcement Class-level Matching Network

更新时间：2026-10-10

- Cross-domain Fault Diagnosis of Gearboxes Based on Reinforcement Class-level Matching Network
- Chinese Journal of Power Engineering Vol. 46, Issue 1, Pages: 148-156(2026)
- 作者机构：
  
  东南大学能源与环境学院, 大型发电装备安全运行与智能测控国家工程研究中心,江苏,南京,210096
- 作者简介：
- 基金信息：
- DOI：10.19805/j.cnki.jcspe.2026.240685
  CLC：
- Online First：15 January 2026，
  
  Published：15 January 2026
- 稿件说明：
移动端阅览
饶朗,邓艾东,邓敏强,吴一凡,夏卫平,胡沁怡. 基于强化类级匹配网络的齿轮箱跨域故障诊断动力工程学报, 2026, 46(1): 148-156 https://doi.

org/10.19805/j.cnki.jcspe.2026.240685
饶朗,邓艾东,邓敏强,吴一凡,夏卫平,胡沁怡. 基于强化类级匹配网络的齿轮箱跨域故障诊断动力工程学报, 2026, 46(1): 148-156 https://doi. DOI： 10.19805/j.cnki.jcspe.2026.240685.

org/10.19805/j.cnki.jcspe.2026.240685 DOI：

摘要

为解决风电机组齿轮箱无监督域适应故障诊断领域存在的分类器决策边界附近样本误分类的问题

提出了一种强化类级匹配网络(RCMN)。该网络在最大化分类器差异的基础上

一方面通过引入基于伪标签的多步训练类级匹配机制

解决决策边界附近样本误分类的问题;另一方面引入自适应对比学习损失函数

解决原有损失函数鲁棒性较差、对参数敏感的缺点。最后

在变转速和变负载2个案例下进行了模型验证。结果表明:在2个案例(共计12个迁移学习任务)中

RCMN的故障诊断准确率均高于最大均值差异、深度相关对齐网络、深度对抗领域适配网络和最大化分类器差异;所提RCMN在解决因分类器决策边界附近样本误分类导致的故障诊断准确率较低的问题上具有一定的有效性和优越性。

Abstract

To address issues of misclassification of samples near the decision boundary of classifier in the field of unsupervised domain adaptation for fault diagnosis of wind turbine gearbox

a reinforcement class-level matching network (RCMN) was proposed. Based on the maximizing classifier discrepancy

on the one hand

a multi-step training class-level matching mechanism based on pseudo-labels was introduced to solve the problem of misclassification of samples near the decision boundary; on the other hand

an adaptive contrastive learning loss function was introduced to address the shortcomings of the original loss function

such as poor robustness and sensitivity to parameters. Finally

model verification was conducted in the two cases of varying speed and varying load. Results show that in a total of 12 transfer learning tasks across the two cases

the fault diagnosis accuracy of RCMN is higher than that of the maximum mean difference

deep correlation alignment network

deep adversarial domain adaptation network

and maximizing classifier discrepancy. The proposed RCMN has certain effectiveness and superiority in addressing the issue of low fault diagnosis accuracy caused by misclassification of samples near the decision boundary of the classifier.

关键词

Keywords

references

ZHU Yunyi, XIE Bin, WANG Anqi, et al. Fault diagnosis of wind turbine gearbox under limited labeled data through temporal predictive and similarity contrast learning embedded with self-attention mechanism[J]. Expert Systems with Applications, 2024, 245: 123080.

SALAMEH J P, CAUET S, ETIEN E, et al. Gearbox condition monitoring in wind turbines: a review[J]. Mechanical Systems and Signal Processing, 2018, 111: 251-264.

LIU Dongdong, CUI Lingli, CHENG Weidong.A review on deep learning in planetary gearbox health state recognition: methods, applications, and dataset publication[J]. Measurement Science and Technology, 2024, 35(1): 012002.

叶柯华, 李春, 胡璇. 基于经验小波变换和关联维数的风力机齿轮箱故障诊断[J]. 动力工程学报, 2021, 41(2): 113-120.YE Kehua, LI Chun, HU Xuan. Fault diagnosis of a wind turbine gearbox based on empirical wavelet transform and correlation dimension[J]. Journal of Chinese Society of Power Engineering, 2021, 41(2): 113-120.

张书瑶, 王梓齐, 刘长良. 基于改进集成KNN回归算法的风电机组齿轮箱状态监测[J]. 动力工程学报, 2023, 43(6): 759-767.ZHANG Shuyao, WANG Ziqi, LIU Changliang. Wind turbine gearbox condition monitoring based on improved ensemble KNN regression algorithm[J]. Journal of Chinese Society of Power Engineering, 2023, 43(6): 759-767.

董路南, 邓艾东, 范永胜, 等. 基于VMD和改进DenseNet的滚动轴承故障诊断[J]. 动力工程学报, 2023, 43(11): 1500-1505, 1522.DONG Lunan, DENG Aidong, FAN Yongsheng, et al. Rolling bearing fault diagnosis based on VMD and improved DenseNet[J]. Journal of Chinese Society of Power Engineering, 2023, 43(11): 1500-1505, 1522.

WANG Weibing, GUO Shuai, ZHAO Shuanfeng, et al. Intelligent fault diagnosis method based on VMD-Hilbert spectrum and ShuffleNet-V2: application to the gears in a mine scraper conveyor gearbox[J]. Sensors, 2023, 23(10): 4951.

WANG Bin, XU Yadong, WANG Manyi, et al. Gear fault diagnosis method based on the optimized graph neural networks[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 3505811.

GE Liang, WANG Yinjun, DING Xiaoxi, et al. A multiscale feature weighted transfer network for unlabeled rotating machinery fault diagnosis[J]. IEEE Sensors Journal, 2024, 24(12): 19305-19315.

苏浩, 杨鑫, 向玲, 等. 基于深度对比迁移学习的变工况下机械故障诊断[J]. 振动工程学报, 2023, 36(3): 845-853.SU Hao, YANG Xin, XIANG Ling, et al. Mechanical fault diagnosis using deep contrastive transfer learning under variable working conditions[J]. Journal of Vibration Engineering, 2023, 36(3): 845-853.

MA Jiaocheng, LV Hongdong, LIU Qin, et al. An unsupervised transfer learning gear fault diagnosis method based on parameter-optimized VMD and residual attention networks[J]. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2024, 46(11): 652.

LI Xueyi, YU Tianyu, WANG Xiangkai, et al. Fusing joint distribution and adversarial networks: a new transfer learning method for intelligent fault diagnosis[J]. Applied Acoustics, 2024, 216: 109767.

SAITO K, WATANABE K, USHIKU Y, et al. Maximum classifier discrepancy for unsupervised domain adaptation[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 3723-3732.

WU Junkang, CHEN Jiawei, WU Jiancan, et al. Understanding contrastive learning via distributionally robust optimization[C]// Proceedings of the 37th International Conference on Neural Information Processing Systems. New Orleans: Curran Associates Inc., 2023: 23297-23320.

马天霆, 孙良海, 韩冰, 等. 基于部分域适应的旋转机械跨工况故障诊断[J]. 太阳能学报, 2024, 45(6): 479-486.MA Tianting, SUN Lianghai, HAN Bing, et al. Cross-working conditions fault diagnosis of rotating machinery based on partial domain adaptation[J]. Acta Energiae Solaris Sinica, 2024, 45(6): 479-486.

LONG Mingsheng, CAO Yue, WANG Jianmin, et al. Learning transferable features with deep adaptation networks[C]//Proceedings of the 32nd International Conference on International Conference on Machine Learning. Lille: JMLR org, 2015: 97-105.

SUN Baochen, SAENK O. Deep CORAL: correlation alignment for deep domain adaptation[C]//Proceedings of European Conference on Computer Vision.Amsterdam: Springer, 2016: 443-450.

GANIN Y, LEMPIITSKY V. Unsupervised domain adaptation by backpropagation[C]//Proceedings of the 32nd International Conference on International Conference on Machine Learning. Lille: JMLR org, 2015: 1180-1189.

VAN DER MAATEN L, HINTON G. Visualizing data using t -SNE[J ] . Journal of Machine Learning Research, 2008, 9(2605): 2579-2605.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on the Application of Singular Value Decomposition in Early Friction Fault Diagnosis

Cross-domain Rotor Fault Diagnosis Method Based on Multi-scale Attention Domain Adaptation Network

Research on Bearing Fault Diagnosis Model and Method Based on Unbalanced Sample Data

Dual-channel Short-term Wind Power Forecasting Fusing Long- and Short-term Dependencies

Research on Trading Decision of Wind-Photovoltaic-Gas Combined System Participating in Day-ahead and Real-time Markets Considering Wind and Photovoltaic Uncertainty

Related Author

CAO Zhiwei

HUANG Jinjun

REN Jie

QIAN Wenhao

YANG Jiangang

XIANG Ling

HAN Zhongquan

BING Hankun

Related Institution

School of Energy and Environment, Southeast University

Electric Power Research Institute of Shandong Electric Power Company, State Grid

Energy and Power Engineering Institute, University of Shanghai for Science and Technology

Huadian Electric Power Research Institute Co., Ltd.

Department of Mechanical Engineering, North China Electric Power University

AI问答

⁰