Meta-class Incremental Intelligent Transfer Diagnostic Method Driven by Multi-source Information Fusion

doi:10.3901/JME.260744

Abstract

Abstract: During the full lifecycle service process of high-end equipment, the continuous emergence of new fault modes and their evolution poses great challenges to lifelong intelligent diagnosis capability of intelligent fault diagnosis models, lifelong intelligent diagnosis technology driven by class-incremental learning has been regarded as a crucial support for ensuring the operational safety of high-end equipment throughout its life cycle. To address the limitations of existing class-incremental learning methods, such as restricted diagnostic performance in class-incremental transfer scenarios, susceptibility to catastrophic forgetting, and insufficient model generalization, proposes a meta-class-incremental intelligent transfer diagnosis method driven by multi-source information fusion. The incorporation of continuous wavelet transform and RGB multi-channel information fusion mechanism is employed to integrate time-frequency features and spatial information from multi-sensor data, enhancing the feature representation capability of fault samples in class-incremental transfer diagnosis scenarios. An enhanced knowledge distillation method is designed and synergized with a memory replay strategy to effectively mitigate the catastrophic forgetting issue in class-incremental transfer diagnosis scenarios. A novel meta-class-incremental parameter learning strategy is developed by incorporating meta-learning into the class-incremental learning framework to effectively improve the model generalization in class-incremental transfer diagnosis scenarios. Finally, the effectiveness and superiority of the proposed method are validated using a planetary transmission fault dataset. Experimental results demonstrate that the proposed method achieves an average diagnostic accuracy of 96.21% and an average forgetting rate of only 1.76% across different class-incremental transfer diagnosis scenarios, thereby verifying its superior generalization ability and strong resistance to forgetting for achieving class-incremental transfer diagnosis. Compared to state-of-the-art methods, the proposed method exhibits advantageous class-incremental transfer diagnostic performance, providing a novel solution for lifelong intelligent fault diagnosis in the health management of high-end equipment throughout its full lifecycle.

Key words: class-incremental learning, meta-learning, knowledge distillation, information fusion, transfer diagnosis

CLC Number:

TH17

KONG Yun, LIN Cuiying, HAN Qinkai, ZHANG Chuntao, DONG Mingming, LIU Hui, CHU Fulei. Meta-class Incremental Intelligent Transfer Diagnostic Method Driven by Multi-source Information Fusion[J]. Journal of Mechanical Engineering, 2026, 62(14): 72-83.

Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks

URL: http://www.cjmenet.com.cn/EN/10.3901/JME.260744

http://www.cjmenet.com.cn/EN/Y2026/V62/I14/72

References

[1] 陈钱，陈康康，董兴建，等. 一种面向机械设备故障诊断的可解释卷积神经网络[J]. 机械工程学报，2024，60(12)：65-76. CHEN Qian，CHEN Kangkang，DONG Xingjian，et al. Interpretable convolutional neural network for mechanical equipment fault diagnosis[J]. Journal of Mechanical Engineering，2024，60(12)：65-76.
[2] 杨诚，何清波，贾民平，等. 流形正则化支持高阶张量机及其在行星齿轮箱半监督故障诊断中的应用[J]. 振动工程学报，2025，38(1)：78-87. YANG Cheng，HE Qingbo，JIA Minping，et al. Manifold regularized support higher-order tensor machines for semi-supervised fault diagnosis of planetary gearboxes[J]. Journal of Vibration Engineering，2025，38(1)：78-87.
[3] 袁静，任港星，蒋会明，等. 基于多元提升核神经网络的机械故障诊断方法及其特征提取可解释性研究[J]. 机械工程学报，2024，60(12)：51-64. YUAN Jing，REN Gangxing，JIANG Huiming，et al. Neural network driven by multiple lifting kernels for mechanical fault diagnosis and its interpretability research of feature extraction[J]. Journal of Mechanical Engineering，2024，60(12)：51-64.
[4] 张洁，孔运，黄国昱，等. 面向高可信机械故障诊断的多模态数据插补与融合方法[J/OL]. 振动工程学报，1-12[2025-04-15]. ZHANG Jie，KONG Yun，HUANG Guoyu，et al. Multimodal data imputation and fusion method for highly trustworthy machine fault diagnosis[J/OL]. Journal of Vibration Engineering，1-12[2025-04-15].
[5] HE H，ZHUANG C，XIONG H. A novel incremental method for bearing fault diagnosis that continuously incorporates unknown fault types[J]. Mechanical Systems and Signal Processing，2024，216：111524.
[6] LIU Y，CHEN B，WANG D，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，2023，72：1-11.
[7] XU X，BAO S，LIANG P，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.
[8] MASANA M，LIU X，TWARDOWSKI B，et al. Class-incremental learning：Survey and performance evaluation on image classification[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence，2023，45(5)：5513-5533.
[9] YAN S，SHAO H，WANG X，et al. Few-shot class-incremental learning for system-level fault diagnosis of wind turbine[J]. IEEE/ASME Transactions on Mechatronics，2025，30：4493-4502.
[10] GU X，ZHAO Y，YANG G，et al. An imbalance modified convolutional neural network with incremental learning for chemical fault diagnosis[J]. IEEE Transactions on Industrial Informatics，2022，18(6)：3630-3639.
[11] SHI M，DING C，CHANG S，et al. Cross-domain class incremental broad network for continuous diagnosis of rotating machinery faults under variable operating conditions[J]. IEEE Transactions on Industrial Informatics，2024，20(4)：6356-6368.
[12] HU K，HE Q，CHENG C，et al. Adaptive incremental diagnosis model for intelligent fault diagnosis with dynamic weight correction[J]. Reliability Engineering & System Safety，2024，241：109705.
[13] WAN X，CEN L，CHEN X，et al. Multiview shapelet prototypical network for few-shot fault incremental learning[J]. IEEE Transactions on Industrial Informatics，2024，20(10)：11751-11762.
[14] XU X，BAO S，SHAO H，et al. A multi-sensor fused incremental broad learning with D-S theory for online fault diagnosis of rotating machinery[J]. Advanced Engineering Informatics，2024，60：102419.
[15] ZHANG Y，HAN D，SHI P. Semi-supervised prototype network based on compact-uniform-sparse representation for rotating machinery few-shot class incremental fault diagnosis[J]. Expert Systems with Applications，2024，255：124660.
[16] ZHANG H，YAO Y，WANG Z，et al. Class incremental fault diagnosis under limited fault data via supervised contrastive knowledge distillation[J]. IEEE Transactions on Industrial Informatics，2025，21(16)：4344-4354.
[17] FINN C，ABBEEL P，LEVINE S. Model-agnostic meta-learning for fast adaptation of deep networks [C]//Proceedings of the 34th International Conference on Machine Learning，Sydney，Australia，Aug 06-11，2017.
[18] REBUFFI S-A，KOLESNIKOV A，SPERL G，et al. iCaRL：Incremental classifier and representation learning [C]//Proceedings of the 30th IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)，Honolulu，HI，Jul 21-26，2017.
[19] HINTON G，VINYALS O，DEAN J J C S. Distilling the knowledge in a neural network[J]. arXiv preprint arXiv：1503.02531，2015.
[20] ZHAO B，XIAO X，GAN G，et al. Maintaining discrimination and fairness in class incremental learning [C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)，Electr Network，Jun 14-19，2020
[21] ZHOU D W，WANG F Y，YE H J，et al. PyCIL：A python toolbox for class-incremental learning[J]. Science China-Information Sciences，2023，66：197101.
[22] YAN S，XIE J，HE X，et al. DER：Dynamically expandable representation for class incremental learning [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)，Electr Network，Jun 19-25，2021.
[23] LIN C，KONG Y，HAN Q，et al. IF-EDAAN：An information fusion-enhanced domain adaptation attention network for unsupervised transfer fault diagnosis[J]. Mechanical Systems and Signal Processing，2025，224：112180.