Multi-Condition Remaining Useful Life Prediction Based on Mixture of Encoders
Abstract
:1. Introduction
- We propose a novel multi-condition RUL prediction framework that dynamically integrates condition-specific and shared knowledge using a Mixture-of-Encoders- and Transformer-based architecture.
- We derive a theoretical lower bound for the error rate of the proposed MoEFormer, providing a formal performance guarantee.
- Through extensive experiments on the C-MAPSS dataset, we demonstrate that MoEFormer outperforms existing state-of-the-art methods in predictive accuracy and robustness.
2. Related Work
2.1. Deep Learning for Sequence Modeling
2.2. Deep Learning for RUL Prediction
3. Preliminary
3.1. Transformer for Sequence Modeling
3.2. RUL Prediction
- is the learned model function;
- represents the model parameters, which are learned during training;
- are the sensor data that describe the system’s state within a window of length L ending at time T.
4. Methods
4.1. Distribution Alignment
4.2. Mixture of Encoders
4.3. Transformer Predictor
5. Theoretical Analysis of MoEFormer
6. Experiments and Results
6.1. Settings
6.1.1. Dataset
6.1.2. Baselines
- CNN [6]: A convolutional-neural-network-based regression approach.
- MODBNE [36]: A multi-objective deep belief network ensemble method.
- Dual-Task LSTM [39]: A dual-task LSTM designed for joint learning of degradation assessment and RUL prediction.
- DCNN [24]: A deep convolutional neural network model that focuses on extracting local data features to improve RUL prediction.
- CNN-LSTM [37]: A hybrid weighted deep domain adaptation approach that combines CNN and LSTM architectures.
- Cap-LSTM [8]: A model that integrates capsule neural networks with LSTM modules for enhanced feature representation.
- BiRNN-ED [40]: An improved similarity-based prognostic model utilizing a bidirectional RNN with an encoder–decoder architecture.
- CNN-BiGRU [27]: A feature-fusion-based method that dynamically adjusts the weights of input features for RUL prediction.
- GCU-Transformer [41]: A Transformer-based architecture incorporating gated convolutional units for improved performance.
- HDNN [42]: A hybrid deep neural network that combines LSTM and CNN layers to extract both temporal and spatial features.
6.1.3. Performance Metrics
6.2. Results
6.3. Ablation Study
6.4. Study on Hyperparameters
7. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
Abbreviations
MoE | Mixture of Encoders |
RUL | Remaining Useful Life |
CNN | Convolutional neural network |
RNN | Recurrent neural network |
LSTM | Long short-term memory |
References
- Si, X.S.; Wang, W.; Hu, C.H.; Zhou, D.H. Remaining useful life estimation—A review on the statistical data driven approaches. Eur. J. Oper. Res. 2011, 213, 1–14. [Google Scholar] [CrossRef]
- Lei, Y.; Li, N.; Guo, L.; Li, N.; Yan, T.; Lin, J. Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mech. Syst. Signal Process. 2018, 104, 799–834. [Google Scholar] [CrossRef]
- Orchard, M.E.; Vachtsevanos, G.J. A particle-filtering approach for on-line fault diagnosis and failure prognosis. Trans. Inst. Meas. Control 2009, 31, 221–246. [Google Scholar] [CrossRef]
- Wang, T.; Yu, J.; Siegel, D.; Lee, J. A similarity-based prognostics approach for remaining useful life estimation of engineered systems. In Proceedings of the International Conference on Prognostics and Health Management, Denver, CO, USA, 6–9 October 2008; pp. 1–6. [Google Scholar]
- Wang, H.; Zhang, X.; Ren, M.; Xu, T.; Lu, C.; Zhao, Z. Remaining useful life prediction of rolling bearings based on multi-scale permutation entropy and ISSA-LSTM. Entropy 2023, 25, 1477. [Google Scholar] [CrossRef] [PubMed]
- Sateesh Babu, G.; Zhao, P.; Li, X.L. Deep convolutional neural network based regression approach for estimation of remaining useful life. In Proceedings of the International Conference on Database Systems for Advanced Applicationsa, Dallas, TX, USA, 16–19 April 2016; Springer: Berlin/Heidelberg, Germany, 2016; pp. 214–228. [Google Scholar]
- Chen, Z.; Wu, M.; Zhao, R.; Guretno, F.; Yan, R.; Li, X. Machine remaining useful life prediction via an attention-based deep learning approach. IEEE Trans. Ind. Electron. 2020, 68, 2521–2531. [Google Scholar] [CrossRef]
- Zhao, C.; Huang, X.; Li, Y.; Li, S. A novel cap-LSTM model for remaining useful life prediction. IEEE Sens. J. 2021, 21, 23498–23509. [Google Scholar] [CrossRef]
- Chang, L.; Lin, Y.h.; Zio, E. Remaining useful life prediction for complex systems considering varying future operational conditions. Qual. Reliab. Eng. Int. 2022, 38, 516–531. [Google Scholar] [CrossRef]
- LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [Google Scholar] [CrossRef]
- Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [Google Scholar] [CrossRef] [PubMed]
- Cho, K.; van Merrienboer, B.; Gulçehre, Ç.; Bahdanau, D.; Bougares, F.; Schwenk, H.; Bengio, Y. Learning phrase representations using RNN encoder—Decoder for statistical machine translation. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, Doha, Qatar, 25–29 October 2014; pp. 1724–1734. [Google Scholar]
- Bai, S.; Kolter, J.Z.; Koltun, V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv 2018, arXiv:1803.01271. [Google Scholar]
- Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A.N.; Kaiser, L.; Polosukhin, I. Attention is all you need. In Proceedings of the International Conference on Neural Information Processing Systems, Long Beach, CA, USA, 4–9 December 2017; pp. 6000–6010. [Google Scholar]
- Devlin, J.; Chang, M.W.; Lee, K.; Toutanova, K. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics, Minneapolis, MN, USA, 2–7 June 2019; pp. 4171–4186. [Google Scholar]
- Brown, T.B.; Mann, B.; Ryder, N.; Subbiah, M.; Kaplan, J.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A.; et al. Language models are few-shot learners. In Proceedings of the International Conference on Neural Information Processing Systems, Vancouver, BC, Canada, 6–12 December 2020. [Google Scholar]
- Malhi, A.; Yan, R.; Gao, R.X. Prognosis of defect propagation based on recurrent neural networks. IEEE Trans. Instrum. Meas. 2011, 60, 703–711. [Google Scholar] [CrossRef]
- Yuan, M.; Wu, Y.; Lin, L. Fault diagnosis and remaining useful life estimation of aero engine using LSTM neural network. In Proceedings of the IEEE International Conference on Aircraft Utility Systems, Beijing, China, 10–12 October 2016; pp. 135–140. [Google Scholar]
- Wu, Y.; Yuan, M.; Dong, S.; Lin, L.; Liu, Y. Remaining useful life estimation of engineered systems using vanilla LSTM neural networks. Neurocomputing 2018, 275, 167–179. [Google Scholar] [CrossRef]
- Zheng, S.; Ristovski, K.; Farahat, A.; Gupta, C. Long short-term memory network for remaining useful life estimation. In Proceedings of the IEEE International Conference on Prognostics and Health Management, Seattle, WA, USA, 11–13 June 2017; pp. 88–95. [Google Scholar]
- Ellefsen, A.L.; Bjørlykhaug, E.; Æsøy, V.; Ushakov, S.; Zhang, H. Remaining useful life predictions for turbofan engine degradation using semi-supervised deep architecture. Reliab. Eng. Syst. Saf. 2019, 183, 240–251. [Google Scholar] [CrossRef]
- Peng, C.; Wu, J.; Wang, Q.; Gui, W.; Tang, Z. Remaining useful life prediction using dual-channel LSTM with time feature and its difference. Entropy 2022, 24, 1818. [Google Scholar] [CrossRef] [PubMed]
- Deng, Y.; Shichang, D.; Shiyao, J.; Chen, Z.; Zhiyuan, X. Prognostic study of ball screws by ensemble data-driven particle filters. J. Manuf. Syst. 2020, 56, 359–372. [Google Scholar] [CrossRef]
- Li, X.; Ding, Q.; Sun, J.Q. Remaining useful life estimation in prognostics using deep convolution neural networks. Reliab. Eng. Syst. Saf. 2018, 172, 1–11. [Google Scholar] [CrossRef]
- Xu, X.; Wu, Q.; Li, X.; Huang, B. Dilated convolution neural network for remaining useful life prediction. J. Comput. Inf. Sci. Eng. 2020, 20, 021004. [Google Scholar] [CrossRef]
- Liu, K.; Shang, Y.; Ouyang, Q.; Widanage, W.D. A data-driven approach with uncertainty quantification for predicting future capacities and remaining useful life of lithium-ion battery. IEEE Trans. Ind. Electron. 2020, 68, 3170–3180. [Google Scholar] [CrossRef]
- Zhu, Q.; Xiong, Q.; Yang, Z.; Yu, Y. A novel feature-fusion-based end-to-end approach for remaining useful life prediction. J. Intell. Manuf. 2023, 34, 3495–3505. [Google Scholar] [CrossRef]
- Xu, Q.; Wang, Z.; Jiang, C.; Jing, Z. Data-driven predictive maintenance framework considering the multi-source information fusion and uncertainty in remaining useful life prediction. Knowl. Based Syst. 2024, 303, 112408. [Google Scholar] [CrossRef]
- Deng, Y.; Du, S.; Wang, D.; Shao, Y.; Huang, D. A calibration-based hybrid transfer learning framework for RUL prediction of rolling bearing across different machines. IEEE Trans. Instrum. Meas. 2023, 72, 3511015. [Google Scholar] [CrossRef]
- Jia, S.; Deng, Y.; Lv, J.; Du, S.; Xie, Z. Joint distribution adaptation with diverse feature aggregation: A new transfer learning framework for bearing diagnosis across different machines. Measurement 2022, 187, 110332. [Google Scholar] [CrossRef]
- Deng, Y.; Lv, J.; Huang, D.; Du, S. Combining the theoretical bound and deep adversarial network for machinery open-set diagnosis transfer. Neurocomputing 2023, 548, 126391. [Google Scholar] [CrossRef]
- Sun, C.; Ma, M.; Zhao, Z.; Tian, S.; Yan, R.; Chen, X. Deep transfer learning based on sparse autoencoder for remaining useful life prediction of tool in manufacturing. IEEE Trans. Ind. Inform. 2018, 15, 2416–2425. [Google Scholar] [CrossRef]
- Cen, Z.; Hu, S.; Hou, Y.; Chen, Z.; Ke, Y. Remaining useful life prediction of machinery based on improved Sample Convolution and Interaction Network. Eng. Appl. Artif. Intell. 2024, 135, 108813. [Google Scholar] [CrossRef]
- Song, L.; Wu, J.; Wang, L.; Chen, G.; Shi, Y.; Liu, Z. Remaining useful life prediction of rolling bearings based on multi-scale attention residual network. Entropy 2023, 25, 798. [Google Scholar] [CrossRef] [PubMed]
- Deng, Y.; Huang, D.; Du, S.; Li, G.; Zhao, C.; Lv, J. A double-layer attention based adversarial network for partial transfer learning in machinery fault diagnosis. Comput. Ind. 2021, 127, 103399. [Google Scholar] [CrossRef]
- Zhang, C.; Lim, P.; Qin, A.K.; Tan, K.C. Multiobjective deep belief networks ensemble for remaining useful life estimation in prognostics. IEEE Trans. Neural Netw. Learn. Syst. 2016, 28, 2306–2318. [Google Scholar] [CrossRef] [PubMed]
- Wu, Z.; Yu, S.; Zhu, X.; Ji, Y.; Pecht, M. A weighted deep domain adaptation method for industrial fault prognostics according to prior distribution of complex working conditions. IEEE Access 2019, 7, 139802–139814. [Google Scholar] [CrossRef]
- Khelif, R.; Chebel-Morello, B.; Malinowski, S.; Laajili, E.; Fnaiech, F.; Zerhouni, N. Direct remaining useful life estimation based on support vector regression. IEEE Trans. Ind. Electron. 2016, 64, 2276–2285. [Google Scholar] [CrossRef]
- Miao, H.; Li, B.; Sun, C.; Liu, J. Joint learning of degradation assessment and RUL prediction for aeroengines via dual-task deep LSTM networks. IEEE Trans. Ind. Inform. 2019, 15, 5023–5032. [Google Scholar] [CrossRef]
- Yu, W.; Kim, I.Y.; Mechefske, C. An improved similarity-based prognostic algorithm for RUL estimation using an RNN autoencoder scheme. Reliab. Eng. Syst. Saf. 2020, 199, 106926. [Google Scholar] [CrossRef]
- Mo, Y.; Wu, Q.; Li, X.; Huang, B. Remaining useful life estimation via transformer encoder enhanced by a gated convolutional unit. J. Intell. Manuf. 2021, 32, 1997–2006. [Google Scholar] [CrossRef]
- Al-Dulaimi, A.; Zabihi, S.; Asif, A.; Mohammadi, A. A multimodal and hybrid deep neural network model for remaining useful life estimation. Comput. Ind. 2019, 108, 186–196. [Google Scholar] [CrossRef]
Model | FD001 | FD002 | FD003 | FD004 | ||||
---|---|---|---|---|---|---|---|---|
(Single-Condition) | (Multi-Condition) | (Single-Condition) | (Multi-Condition) | |||||
RMSE | Scorex↓ | RMSE | Score↓ | RMSE | Score↓ | RMSE | Score↓ | |
CNN | 18.45 | 1286.7 | 30.29 | 13,570 | 19.82 | 1596.2 | 29.16 | 7886.4 |
MODBNE | 15.04 | 334.2 | 25.05 | 5585.3 | 12.51 | 421.9 | 28.66 | 6557.6 |
Dual-Task LSTM | 12.29 | - | 17.87 | - | 14.34 | - | 21.81 | - |
DCNN | 12.61 | 273.7 | 22.36 | 10,412 | 12.64 | 284.1 | 23.31 | 12,466 |
CNN-LSTM | 14.40 | 290.0 | 27.36 | 9869.0 | 14.32 | 316.0 | 26.69 | 6594.0 |
Cap-LSTM | 12.27 | 260.0 | 17.19 | 1850.0 | 12.55 | 217.0 | 22.05 | 4570.0 |
BiRNN-ED | 13.58 | 228.0 | 19.59 | 2650.0 | 19.16 | 1727.0 | 22.15 | 2901.0 |
CNN-BiGRU | 12.31 | 252.1 | 16.06 | 1238.1 | 12.37 | 283.5 | 19.83 | 2706.8 |
GCU-Transformer | 11.27 | - | 22.81 | - | 11.42 | - | 24.86 | - |
HDNN | 13.02 | 245.0 | 15.24 | 1282.4 | 12.22 | 287.7 | 18.16 | 1527.4 |
MoEFormer | 15.13 | 395.60 | 13.71 | 792.64 | 13.88 | 290.67 | 14.47 | 1008.25 |
Model | FD002 | FD004 | ||
---|---|---|---|---|
RMSE | Score↓ | RMSE | Score↓ | |
MoEFormer w/o DA&MoE | 30.01 | 34,464.83 | 33.99 | 89,148.18 |
MoEFormer w/o DA | 20.76 | 2450.17 | 28.53 | 10,390.44 |
MoEFormer w/o MoE | 24.08 | 6267.54 | 23.60 | 4763.35 |
MoEFormer | 13.71 | 792.64 | 14.47 | 1008.25 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Liu, Y.; Xu, B.; Geng, Y.-a. Multi-Condition Remaining Useful Life Prediction Based on Mixture of Encoders. Entropy 2025, 27, 79. https://doi.org/10.3390/e27010079
Liu Y, Xu B, Geng Y-a. Multi-Condition Remaining Useful Life Prediction Based on Mixture of Encoders. Entropy. 2025; 27(1):79. https://doi.org/10.3390/e27010079
Chicago/Turabian StyleLiu, Yang, Bihe Xu, and Yangli-ao Geng. 2025. "Multi-Condition Remaining Useful Life Prediction Based on Mixture of Encoders" Entropy 27, no. 1: 79. https://doi.org/10.3390/e27010079
APA StyleLiu, Y., Xu, B., & Geng, Y.-a. (2025). Multi-Condition Remaining Useful Life Prediction Based on Mixture of Encoders. Entropy, 27(1), 79. https://doi.org/10.3390/e27010079