The Remaining Useful Life Prediction of Bearings Based on ICPO-TCN

Jiayu Tian, Tiantian Liang, Zhuangzhuang Ma
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:278-288, 2025.

Abstract

In modern industrial equipment maintenance management, bearings, as key rotating components, perform a crucial role in maintaining the stable performance of machinery. Therefore, this paper proposes a bearing Remaining Useful Life (RUL) prediction method based on the Improved Crested Porcupine Optimizer-Time Convolutional Network (ICPO-TCN). Firstly, the improved crested porcupine optimizer (ICPO) is used to search for the best number of modes and penalty factor in VMD, enabling the selection of effective components for signal reconstruction, noise reduction, and enhanced time-frequency feature extraction. A feature dataset is then constructed by combining the selected time-domain and frequency-domain characteristics. Next, reducing the dimensionality by kernel principal component analysis (KPCA), which is then used as input for the TCN model. Finally, ICPO is again employed to optimize the convolution kernel size and learning rate of the TCN to improve RUL prediction accuracy. Experimental results demonstrate that ICPO-TCN outperforms traditional TCN and LSTM models, achieving higher prediction accuracy.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-tian25b, title = {The Remaining Useful Life Prediction of Bearings Based on ICPO-TCN}, author = {Tian, Jiayu and Liang, Tiantian and Ma, Zhuangzhuang}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {278--288}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/tian25b/tian25b.pdf}, url = {https://proceedings.mlr.press/v278/tian25b.html}, abstract = {In modern industrial equipment maintenance management, bearings, as key rotating components, perform a crucial role in maintaining the stable performance of machinery. Therefore, this paper proposes a bearing Remaining Useful Life (RUL) prediction method based on the Improved Crested Porcupine Optimizer-Time Convolutional Network (ICPO-TCN). Firstly, the improved crested porcupine optimizer (ICPO) is used to search for the best number of modes and penalty factor in VMD, enabling the selection of effective components for signal reconstruction, noise reduction, and enhanced time-frequency feature extraction. A feature dataset is then constructed by combining the selected time-domain and frequency-domain characteristics. Next, reducing the dimensionality by kernel principal component analysis (KPCA), which is then used as input for the TCN model. Finally, ICPO is again employed to optimize the convolution kernel size and learning rate of the TCN to improve RUL prediction accuracy. Experimental results demonstrate that ICPO-TCN outperforms traditional TCN and LSTM models, achieving higher prediction accuracy.} }
Endnote
%0 Conference Paper %T The Remaining Useful Life Prediction of Bearings Based on ICPO-TCN %A Jiayu Tian %A Tiantian Liang %A Zhuangzhuang Ma %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-tian25b %I PMLR %P 278--288 %U https://proceedings.mlr.press/v278/tian25b.html %V 278 %X In modern industrial equipment maintenance management, bearings, as key rotating components, perform a crucial role in maintaining the stable performance of machinery. Therefore, this paper proposes a bearing Remaining Useful Life (RUL) prediction method based on the Improved Crested Porcupine Optimizer-Time Convolutional Network (ICPO-TCN). Firstly, the improved crested porcupine optimizer (ICPO) is used to search for the best number of modes and penalty factor in VMD, enabling the selection of effective components for signal reconstruction, noise reduction, and enhanced time-frequency feature extraction. A feature dataset is then constructed by combining the selected time-domain and frequency-domain characteristics. Next, reducing the dimensionality by kernel principal component analysis (KPCA), which is then used as input for the TCN model. Finally, ICPO is again employed to optimize the convolution kernel size and learning rate of the TCN to improve RUL prediction accuracy. Experimental results demonstrate that ICPO-TCN outperforms traditional TCN and LSTM models, achieving higher prediction accuracy.
APA
Tian, J., Liang, T. & Ma, Z.. (2025). The Remaining Useful Life Prediction of Bearings Based on ICPO-TCN. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:278-288 Available from https://proceedings.mlr.press/v278/tian25b.html.

Related Material