Article Details

Abstract

The rolling bearing is a key component of rotating machinery in power plants. Accurate prediction of its Remaining Useful Life (RUL) is very important for maintenance management and fault prevention. In order to improve the prediction accuracy of residual life of rolling bearings, a Deep Learning model combining Convolutional Denoising Autoencoder (CDAE), One-Dimensional Convolutional Neural Network (1D-CNN) and Bidirectional Long Short-Term Memory network (BiLSTM) is proposed. The CDAE module improves data quality by reducing noise in vibration signals, 1D-CNN captures local time features, and BiLSTM effectively models long-term dependencies in time series. In addition, a custom weighted mean square error loss function is introduced to highlight the key stages of the time series, and higher weights are given to the early degradation stage and the late failure stage. It effectively improves the accuracy of the model in the remaining life prediction, especially in the key degradation stage. The verification results on the bearing data set show that the proposed model MAE is only 0.078 6 and RMSE is only 0.096 1, which has better prediction accuracy. The prediction performance of the proposed model is better than that of the existing methods.

Keywords

rolling bearing; Remaining Useful Life; Deep Learning model; CDAE; Bidirectional Long Short-Term Memory network

References

1. [1] HAKIM M，OMRAN A A B，AHMED A N，et al. A systematic review of Rolling Bearing Fault Diagnoses Based on Deep Learning and Transfer Learning: Taxonomy, Overview, Application, Open Challenges, Weaknesses and Recommendations [J/OL].Ain Shams Engineering Journal，2023，14（4）：101945（2023-04-05）[2025-10-09].https://doi.org/10.1016/j.asej.2022.101945.
2. [2] Huang Yaxin, Li Yanting Remaining life prediction of rolling bearings based on dual domain convolutional transformer [J]. Machine Tool and Hydraulic, 2025, 53 (20): 1-7
3. [3] CARTIAUX F-B，EHRLACHER A，LEGOLL F，et al. Probabilistic Formulation of Miner's Rule and Application to Structural Fatigue [J/OL].arXiv:2303.14504 [cs.CE].（2023-03-25）.https://arxiv.org/abs/2303.14504.
4. [4] Li Junxing, Fan Jiahui, Wang Zhihua, etc Life prediction method for rolling bearings under noisy environment based on IKF-ARIMA-NARNN [J]. Journal of Aerodynamics, 2025, 40 (6): 245-253
5. [5] Lei Wenping, Zou Dongliang, Chen Shijin, etc Bayesian Change Point Detection for Remaining Life Prediction of Rolling Bearings [J]. Journal of Zhengzhou University (Engineering Edition), 2025, 46 (6): 93-101
6. [6] MENG Z，QIN X Y，LIU J B，et al. A Denoising Algorithm Based on ARIMA and Competitive K-SVD for the Diagnosis of Rolling Bearing Faults [J/OL].Applied Acoustics，2025，228：110309（2025-01-15）[2025-10-11].https://doi.org/10.1016/j.apacoust.2024.110309.
7. [7] Quanhang, Zhang Qiang, Shao Siyu, etc Remaining life prediction of rolling bearings based on CNN WaveNet [J]. Computer Application Research, 2021, 38 (10): 3098-3103
8. [8] Shen Yanbin, Zhang Xiaoli, Xia Yong, etc Research on Bi LSTM neural network for predicting the remaining service life of bearings [J]. Journal of Vibration Engineering, 2021, 34 (2): 411-420
9. [9] GURINDER K，FEI L，YI P P C. The Integration of Knowledge Graph Convolution Network with Denoising Autoencoder [J].Engineering Applications of Artificial Intelligence，2024，135：108792[2025-10-11].https://doi.org/10.1016/j.engappai.2024.108792.
10. [10] DONG S J，XIAO J F，HU X L，et al. Deep Transfer Learning Based on Bi-LSTM and Attention for Remaining Useful Life Prediction of Rolling Bearing [J/OL]. Reliability Engineering & System Safety，2023，230：108914[2025-10-11].https://doi.org/10.1016/j.ress.2022.108914.

About the authors

Zhang Lin (1980-), male, Han ethnicity, from Renshou, Sichuan, is an engineer with a bachelor's degree. His research interests include thermal automation, intelligent power plant construction, and economic management.