针对风电机组滚动轴承在变工况下存在数据分布偏移及未知故障识别困难的问题,提出一种基于多分类器开放对抗网络（MCOAN）的开放集故障诊断方法。首先,在对抗域适应框架中设置K元和K+1元分类器,分别计算目标样本与源域的相似度;然后,利用相似度构建动态权重机制,实现目标样本在开放集对抗训练过程中的自适应加权,同时提供已知和未知分类的动态阈值,促进已知类特征的跨域分布对齐,提高未知样本的识别精度。此外,引入非对抗分类器以提升动态权重计算准确性。最后,在两个数据集上进行实验,结果表明所提方法能够实现高精度的共享类特征分布对齐和未知类识别,且具有较好的鲁棒性。
Abstract
To address the challenges of data distribution shift and unknown fault identification in wind turbine rolling bearings under varying operating conditions, an open-set fault diagnosis method based on a multi-classifier open adversarial network （MCOAN） is proposed. Specifically, a K-class and a （K+1）-class classifier are jointly employed within an adversarial domain adaptation framework to evaluate the similarity between target and source domain samples. Based on these similarity measures, a dynamic weighting mechanism is constructed to achieve adaptive sample weighting during open-set adversarial training. This mechanism also facilitates the dynamic determination of decision thresholds for known and unknown classes, thereby promoting cross-domain alignment of shared features while enhancing the recognition of unknown samples. In addition, a non-adversarial classifier is introduced to improve the reliability of dynamic weight computation. Experimental results on two datasets demonstrate that the proposed method achieves accurate alignment of shared features and robust identification of unknown faults, showcasing superior diagnostic performance and robustness.
关键词
风电机组 /
故障诊断 /
滚动轴承 /
对抗训练 /
域适应 /
开放集
Key words
wind turbines /
fault diagnosis /
rolling bearings /
adversarial training /
domain adaptation /
open set
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 周忠志, 邓艾东, 刘东瀛, 等. 基于穿越可视图和图同构网络的风电传动系统故障诊断方法[J]. 太阳能学报, 2025, 46(2): 591-599.
Zhou Z Z, Deng A D, Liu D Y, et al.Fault diagnosis method for wind power transmission system based on penetrable visibility graph and graph isomorphism networks[J]. Acta Energiae Solaris Sinica, 2025, 46(2): 591-599.
[2] 刘俊孚, 岑健, 黄汉坤, 等. 零小样本旋转机械故障诊断综述[J]. 计算机工程与应用, 2024, 60(15): 42-54.
Liu J F, Cen J, Huang H K, et al.Review on zero or few sample rotating machinery fault diagnosis[J]. Computer Engineering and Applications, 2024, 60(15): 42-54.
[3] 史宗辉, 陈长征, 安文杰, 等. 基于SVDS-MSCNN的风电机组滚动轴承故障诊断[J]. 振动、测试与诊断, 2024, 44(6): 1152-1158.
Shi Z H, Chen C Z, An W J, et al.Fault diagnosis of wind turbine rolling bearing based on SVDS-MSCNN[J]. Journal of Vibration, Measurement & Diagnosis, 2024, 44(6): 1152-1158.
[4] 李继猛, 王泽, 史清心, 等. 基于图正则化约束频域组稀疏模型的风电机组滚动轴承故障诊断[J]. 中国机械工程, 2024, 35(11): 1909-1919.
Li J M, Wang Z, Shi Q X, et al.Rolling bearing fault diagnosis of wind turbines based on frequency domain group sparse model with graph regularization constraints[J]. China Mechanical Engineering, 2024, 35(11): 1909-1919.
[5] 刘洋, 程强, 史曜炜, 等. 基于注意力模块及1D-CNN的滚动轴承故障诊断[J]. 太阳能学报, 2022, 43(3): 462-468.
Liu Y, Cheng Q, Shi Y W, et al.Fault diagnosis of rolling bearings based on attention module and 1D-CNN[J]. Acta Energiae Solaris Sinica, 2022, 43(3): 462-468.
[6] Ge Y, Zhang F S, Ren Y.Adaptive fault diagnosis method for rotating machinery with unknown faults under multiple working conditions[J]. Journal of Manufacturing Systems, 2022, 63: 177-184.
[7] 安文杰, 陈长征, 田淼, 等. 基于迁移学习的风电机组轴承故障诊断研究[J]. 太阳能学报, 2023, 44(6): 367-373.
An W J, Chen C Z, Tian M, et al.Research on bearing fault diagnosis of wind turbines based on transfer learning[J]. Acta Energiae Solaris Sinica, 2023, 44(6): 367-373.
[8] Liu Y, Deng A D, Deng M Q, et al.Transforming the open set into a pseudo-closed set: a regularized GAN for domain adaptation in open-set fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 3531312.
[9] Ganin Y, Lempitsky V S.Unsupervised domain adaptation by backpropagation[C]//Proceedings of the 32nd International Conference on Machine Learning. Lille, France, 2015.
[10] Xu Z W, Han G J, Chen C L, et al.A novel clustering based on consensus knowledge for cross-domain fault diagnoses[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 3509010.
[11] Jia S Y, Deng Y F, Lyu J, et al.Joint distribution adaptation with diverse feature aggregation: a new transfer learning framework for bearing diagnosis across different machines[J]. Measurement, 2022, 187: 110332.
[12] Xu H Y, Peng X Y, Wang J L, et al.Adaptive graph-guided joint soft clustering and distribution alignment for cross-load and cross-device rotating machinery fault transfer diagnosis[J]. Measurement Science and Technology, 2024, 35(4): 045009.
[13] 李灿, 王广斌, 赵树标, 等. 采用集约化开放集差异分布对齐策略的轴承故障诊断方法[J]. 中国机械工程, 2024, 35(9): 1622-1633.
Li C, Wang G B, Zhao S B, et al.Bearing fault diagnosis method based on intensive distribution alignment with open set difference[J]. China Mechanical Engineering, 2024, 35(9): 1622-1633.
[14] Saito K, Yamamoto S, Ushiku Y, et al.Open set domain adaptation by backpropagation[C]//Computer Vision- ECCV 2018. Cham: Springer, 2018: 156-171.
[15] Yu X L, Zhao Z B, Zhang X W, et al.Deep-learning-based open set fault diagnosis by extreme value theory[J]. IEEE Transactions on Industrial Informatics, 2022, 18(1): 185-196.
[16] Han T, Li Y F.Out-of-distribution detection-assisted trustworthy machinery fault diagnosis approach with uncertainty-aware deep ensembles[J]. Reliability Engineering & System Safety, 2022, 226: 108648.
[17] Zhang Y C, Ji J C, Ren Z H, et al.Multi-sensor open-set cross-domain intelligent diagnostics for rotating machinery under variable operating conditions[J]. Mechanical Systems and Signal Processing, 2023, 191: 110172.
[18] 范苍宁, 刘鹏, 肖婷, 等. 深度域适应综述: 一般情况与复杂情况[J]. 自动化学报, 2021, 47(3): 515-548.
Fan C N, Liu P, Xiao T, et al.A review of deep domain adaptation: general situation and complex situation[J]. Acta Automatica Sinica, 2021, 47(3): 515-548.
基金
江苏省碳达峰碳中和科技创新专项资金（BT2024004; BE2023854）; 中央高校基本科研业务费专项（2242024k30046; 2242024k30047)
PDF(2286 KB)
