In real-world industrial scenarios, the rapid evolution of new fault patterns under complex operating conditions imposes great challenges for class-incremental learning (CIL), such as catastrophic forgetting of previously learned knowledge and poor generalization caused by overfitting under limited data. To address these issues, this study proposes an intelligent diagnostic framework based on pseudo-incremental learning and dual-model coordination. The framework first constructs a generalizable feature representation space through pretraining a base model on fundamental fault classes. A perturbation-based sample augmentation strategy is then employed to generate a structured dataset. A dual-objective optimization framework, which incorporates coordination tasks (between old and new classes) and specialized training tasks for new classes, is then used to enable effective integration of evolving fault modes. Finally, a lightweight parameter fine-tuning mechanism is applied. Experimental results on multiple bearing datasets demonstrate that the proposed method exhibits excellent dynamic adaptability and robustness in few-shot class-incremental learning (FSCIL) scenarios.
GongXFengKDuW, et al. An imbalance multi-faults data transfer learning diagnosis method based on finite element simulation optimization model of rolling bearing. Proc Inst Mech Eng C J Mech Eng Sci2024; 238(17): 8924–8940.
2.
HayesTLKafleKShresthaR, et al. Remind your neural network to prevent catastrophic forgetting. In: European conference on computer vision, Glasgow, UK, 2020, pp.466–483. Springer.
3.
ZhouDWYeHJMaL, et al. Few-shot class-incremental learning by sampling multi-phase tasks. IEEE Trans Pattern Anal Mach Intell2023; 45(11): 12816–12831.
4.
LiuHGuLChiZ, et al. Few-shot class-incremental learning via entropy-regularized data-free replay. In: European conference on computer vision, Tel Aviv, Israel, 2022, pp.146–162. Springer.
5.
WangFYZhouDWYeHJ, et al. Foster: feature boosting and compression for class-incremental learning. In: European conference on computer vision, Tel Aviv, Israel, 2022, pp.398–414. Springer.
6.
ZhouDWWangFYYeHJ, et al. Forward compatible few-shot class-incremental learning. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, New Orleans, LA, USA, 2022, pp.9046–9056.
7.
XieSShenCWangD, et al. A new lifelong learning method based on dual distillation for bearing diagnosis with incremental fault types. Adv Eng Inform2025; 65: 103136.
8.
LiuRDiaoBHuangL, et al. Low-redundancy distillation for continual learning. Pattern Recognit2025; 167: 111712.
9.
LiuLJiJZhaoL. Befm: a balanced and efficient fine-tuning model in class-incremental learning. Knowl Syst2025; 315(000): 113298.
10.
LiZHoiemD. Learning without forgetting. IEEE Trans Pattern Anal Mach Intell2018; 40(12): 2935–2947.
11.
KirkpatrickJPascanuRRabinowitzN, et al. Overcoming catastrophic forgetting in neural networks. Proc Natl Acad Sci2017; 114(13): 3521–3526.
12.
RannenAAljundiRBlaschkoMB, et al. Encoder based lifelong learning. In: Proceedings of the IEEE international conference on computer vision, Venice, Italy, 2017, pp.1320–1328.
13.
MallyaALazebnikS.Packnet: adding multiple tasks to a single network by iterative pruning. In: Proceedings of the IEEE conference on computer vision and pattern recognition, Salt Lake City, UT, USA, 2018, pp.7765–7773.
14.
ZhaoCShenW. Dual adversarial network for cross-domain open set fault diagnosis. Reliab Eng Syst Saf2022; 221: 108358.
15.
SuHXiangLHuA, et al. A novel method based on meta-learning for bearing fault diagnosis with small sample learning under different working conditions. Mech Syst Signal Process2022; 169: 108765.
16.
LiJLiuTWuX. Research on bearing vibration signal generation method based on filtering wgan_gp with small samples. Proc Inst Mech Eng C J Mech Eng Sci2023; 237(20): 4911–4929.
17.
CheCWangHLinR, et al. Deep meta-learning and variational autoencoder for coupling fault diagnosis of rolling bearing under variable working conditions. Proc Inst Mech Eng C J Mech Eng Sci2022; 236(17): 9900–9913.
18.
CheCWangHXiongM, et al. Few-shot fault diagnosis of rolling bearing under variable working conditions based on ensemble meta-learning. Digit Signal Process2022; 131: 103777.
19.
LiJLiuTWuX.Research on bearing vibration signal generation method based on filtering WGAN_GP with small samples. Proc Inst Mech Eng Part C: J Mech Eng Sci2023; 237(20): 4911–4929.
20.
ZhuJYaoGZhangW, et al. Feature distribution distillation-based few shot class incremental learning. In: Proceedings of the 5th International Conference on Pattern Recognition and Artificial Intelligence (PRAI), Chengdu, China, 2022, pp.108–113.
21.
JiaoLWangDBaiY, et al. Deep learning in visual tracking: a review. IEEE Trans Neural Netw Learn Syst2023; 34(9): 5497–5516.
22.
ShankarampetaARYamauchiK.Few-shot class incremental learning with generative feature replay. In: Proceedings of the 10th international conference on pattern recognition applications and methods (ICPRAM), Virtual conference, 2021, pp.259–267.
23.
WangYWangYZhaoG, et al. Learning to complement: relation complementation network for few-shot class-incremental learning. Knowl Syst2023; 282: 111130.
24.
ZhuHShenCWangJ, et al. Few-shot class-incremental learning with adjustable pseudo-incremental sessions for bearing fault diagnosis. IEEE Sens J2024; 24: 19543–19552.
25.
YangBLinMZhangY, et al. Dynamic support network for few-shot class incremental learning. IEEE Trans Pattern Anal Mach Intell2022; 45(3): 1–2951.
26.
TaoXHongXChangX, et al. Few-shot class-incremental learning. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, Seattle, WA, USA (Fully Virtual), 2020, pp.12183–12192.
27.
PengCKoniuszPGuoK, et al. Multivariate prototype representation for domain-generalized incremental learning. Comput Vis Image Underst2024; 249: 104215.
28.
HuKBiZHeQ, et al. A feature extension and reconstruction method with incremental learning capabilities under limited samples for intelligent diagnosis. Adv Eng Inform2024; 62: 102796.
29.
LiPLiuFJiaoL, et al. Task context transformer and GCN for few-shot learning of cross-domain. Neurocomputing2023; 548: 126433.
30.
ZhaoLLuJXuY, et al. Few-shot class-incremental learning via class-aware bilateral distillation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, Vancouver, BC, Canada, 2023, pp.11838–11847.
31.
RebuffiSAKolesnikovASperlG, et al. ICARL: incremental classifier|representation learning. In: Proceedings of the IEEE conference on computer vision and pattern recognition, Honolulu, HI, USA, 2017, pp.2001–2010.
32.
HouSPanXLoyCC, et al. Learning a unified classifier incrementally via rebalancing. In: 2019 IEEE/CVF conference on computer vision and pattern recognition (CVPR), Long Beach, CA, USA, 2019, pp.831–839.
33.
ZhangCSongNLinG, et al. Few-shot incremental learning with continually evolved classifiers. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, Nashville, TN, USA, 2021, pp.12455–12464.
