面向小样本场景的风机多增量故障诊断

doi:10.3901/JME.2025.10.230

摘要/Abstract

摘要： 作为一种复杂工业设备，风机在长期运行过程中将不可避免出现新的故障。增量故障诊断能从连续的数据流中不断积累新的故障知识，以持续扩展模型的诊断能力，并克服灾难性遗忘。然而，在面向风机复杂的多增量阶段与小故障样本的场景时，增量诊断模型仍会出现“稳定性-可塑性”的困境。因此，提出一种动态记忆策略自驱动的增量故障诊断方法，在故障样本有限的场景下，实现多增量阶段的风机智能诊断。首先，构造自适应样例库模块，通过自主地提炼存储具有代表性的旧类别样本，以避免多增量诊断过程的灾难性遗忘。其次，设计权值动态校正算法，帮助模型实时修正各类别节点之间的重要程度，以灵活地学习有限故障数据中的诊断知识。在风机的小故障样本多增量阶段的诊断试验中，通过与当前主流的增量学习方法多维对比，验证了所提方法的优越性。

关键词: 多增量故障诊断, 小样本, 风力发电机, 自适应样例库, 权值动态修正算法

Abstract: As a complex industrial device, wind turbines will inevitably develop new faults during prolonged operation. Incremental fault diagnosis can continuously accumulate new fault knowledge from ongoing data streams, thus expanding diagnostic capabilities of the model and overcoming catastrophic forgetting. However, in scenarios with complex multi-incremental stages and small fault samples for wind turbines, the incremental diagnostic model still faces the “stability-plasticity” dilemma. Therefore, a dynamic memory strategy-driven incremental fault diagnosis method is proposed to achieve intelligent diagnosis of wind turbines in multi-increment stages under limited fault sample scenarios. First, an adaptive example repository module is constructed, autonomously refining and storing representative samples of previous categories to prevent catastrophic forgetting during the multi-incremental diagnostic process. Secondly, a weight dynamic correction algorithm is designed to help the model real-time adjust the importance between different category nodes, facilitating flexible learning of diagnostic knowledge from limited fault data. In the diagnostic experiments during the multi-incremental stages of small fault samples in the wind turbine, the superiority of the proposed method is validated by the multidimensional comparison with current mainstream incremental learning methods.

Key words: multi-incremental fault diagnosis, small samples, wind turbine, adaptive example repository, weight dynamic correction algorithm

中图分类号:

TH17

邵海东, 颜深, 刘政武, 肖一鸣, 韩特. 面向小样本场景的风机多增量故障诊断[J]. 机械工程学报, 2025, 61(10): 230-240.

SHAO Haidong, YAN Shen, LIU Zhengwu, XIAO Yiming, HAN Te. Multi-incremental Fault Diagnosis of Wind Turbine Towards Small Sample Scenario[J]. Journal of Mechanical Engineering, 2025, 61(10): 230-240.

参考文献

[1] HAN Te，XIE Wenzhen，PEI Zhongyi. Semi-supervised adversarial discriminative learning approach for intelligent fault diagnosis of wind turbine[J]. Information Sciences，2023，648：119496.
[2] LIU Guanglun，SI Jianlong，MENG Wenchao，et al. Wind turbine fault detection with multimodule feature extraction network and adaptive strategy[J]. IEEE Transactions on Instrumentation and Measurement，2023，72：1-13.
[3] 雷亚国，杨彬，李乃鹏，等. 跨设备的机械故障靶向迁移诊断方法[J]. 机械工程学报，2022，58(12)：1-9.LEI Yaguo，YANG Bin，LI Naipeng，et al. Targeted transfer diagnosis method across different machines[J]. Journal of Mechanical Engineering，2022，58(12)：1-9.
[4] 邵海东，肖一鸣，颜深. 仿真数据驱动的改进无监督域适应轴承故障诊断[J]. 机械工程学报，2023，59(3)：76-85.SHAO Haidong，XIAO Yiming，YAN Shen. Simulation data-driven enhanced unsupervised domain adaptation for bearing fault diagnosis[J]. Journal of Mechanical Engineering，2023，59(3)：76-85.
[5] SHANG Zuogang，ZHAO Zhibin，YAN Ruqiang. Denoising fault-aware wavelet network：A signal processing informed neural network for fault diagnosis[J]. Chinese Journal of Mechanical Engineering，2023，36(1)：9.
[6] 邵海东，肖一鸣，闵志闪，等. 区块链和边缘计算赋能的联邦学习故障诊断框架[J]. 机械工程学报，2023，59(21)：283-292.SHAO Haidong，XIAO Yiming，MIN Zhishan，et al. Blockchain and edge computing enabled federated learning fault diagnosis framework[J]. Journal of Mechanical Engineering，2023，59(21)：283-292.
[7] DING Ao，YI Xiaojian，QIN Yong，et al. Self-driven continual learning for class-added motor fault diagnosis based on unseen fault detector and propensity distillation[J]. Engineering Applications of Artificial Intelligence，2024，127：107382.
[8] LIU Yao，CHEN Bojian，WANG Dong，et al. A lifelong learning method based on generative feature replay for bearing diagnosis with incremental fault types[J]. IEEE Transactions on Instrumentation and Measurement，2023，72：3514311.
[9] CHEN Bojian，SHEN Changqing，SHI Juanjuan，et al. Continual learning fault diagnosis：A dual-branch adaptive aggregation residual network for fault diagnosis with machine increments[J]. Chinese Journal of Aeronautics，2023，36(6)：361-377.
[10] HU Kui，HE Qingbo，CHENG Changming，et al. Adaptive incremental diagnosis model for intelligent fault diagnosis with dynamic weight correction[J]. Reliability Engineering & System Safety，2024，241：109705.
[11] LI Chuan，LEI Xiaotong，HUANG Yunwei，et al. Incrementally contrastive learning of homologous and interclass features for the fault diagnosis of rolling element bearings[J]. IEEE Transactions on Industrial Informatics，2023，19：11182-11191.
[12] XU Xuefang，BAO Shuo，LIANG Pengfei，et al. A broad learning model guided by global and local receptive causal features for online incremental machinery fault diagnosis[J]. Expert Systems with Applications，2024，246：123124.
[13] LI Jipu，HUANG Ruyi，CHEN Zhuyun，et al. Deep continual transfer learning with dynamic weight aggregation for fault diagnosis of industrial streaming data under varying working conditions[J]. Advanced Engineering Informatics，2023，55：101883.
[14] SHI Mingkuan，DING Chuancang，WANG Rui，et al. Semi-supervised class incremental broad network for continuous diagnosis of rotating machinery faults with limited labeled samples[J]. Knowledge-Based Systems，2024，286：111397.
[15] MASANA M，LIU Xialei，TWARDOWSKI B，et al. Class-incremental learning：Survey and performance evaluation on image classification[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence，2022，45(5)：5513-5533.
[16] KIRKPATRICK J，PASCANU R，RABINOWITZ N，et al. Overcoming catastrophic forgetting in neural networks[J]. Proceedings of the National Academy of Sciences，2017，114(13)：3521-3526.
[17] BUZZEGA P，BOSCHINI M，PORRELLO A，et al. Dark experience for general continual learning：A strong，simple baseline[J]. Advances in Neural Information Processing Systems，2020，33：15920-15930.
[18] REBUFFI S A，KOLESNIKOV A，SPERL G，et al. iCaRL：Incremental classifier and representation learning[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)，Honolulu，HI，USA，2017：2001-2010.
[19] ATKINSON C，MCCANE B，SZYMANSKI L，et al. Pseudo-rehearsal：Achieving deep reinforcement learning without catastrophic forgetting[J]. Neurocomputing，2021，428：291-307.
[20] LOPEZ-PAZ D，RANZATO M A. Gradient episodic memory for continual learning[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems，December 4 - 9，2017，Long Beach California，USA. NewYork：Curran Associates Inc， 2017：6470 - 6479.
[21] CHAUDHRY A，RANZATO M A，ROHRBACH M，et al. Efficient lifelong learning with A-GEM[C]// The Seventh International Conference on Learning Representations，New Orleans，Louisiana，United States，May 6-9，2019.
[22] GUAN Liuen，QIAO Fei，ZHAI Xiaodong，et al. Model evolution mechanism for incremental fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement，2022，71：3522111.
[23] 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)，Las Vegas，NV，USA，2016：770-778.
[24] PAN S J，YANG Qiang. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering，2010，22(10)：1345-1359.