Research on Machine Tool Bearing Compound Fault Recognition Based on Adaptive Wavelet Threshold Denoising and Improved CNN-BiLSTM
Main Article Content
Abstract
As the core supporting component of machine tools, the operating state of rolling bearings directly determines the machining accuracy and equipment stability of CNC machine tools. Compound faults suffer from strong coupling interference, high noise mixing degree and difficult feature extraction, which lead to low accuracy and poor robustness of traditional fault recognition methods. To address this problem, this paper proposes a machine tool bearing compound fault recognition method integrating adaptive wavelet threshold denoising and improved CNN-BiLSTM. Firstly, aiming at the defects of signal distortion and insufficient denoising existing in traditional wavelet soft and hard threshold denoising, an adaptive threshold function is constructed to dynamically adjust the threshold and wavelet decomposition layers according to the signal noise level, so as to realize efficient denoising of bearing vibration signals. Secondly, multi-scale convolution kernels and dual-channel attention mechanisms are introduced on the basis of the basic CNN-BiLSTM model to simultaneously mine spatial features and time-series correlation features of vibration signals, adapting to the coupling characteristics of compound fault features. Finally, model training and verification are completed based on the measured bearing fault dataset. Experimental results show that compared with traditional deep learning models, the proposed improved model achieves better denoising performance, and the recognition accuracy of compound faults reaches 98.72%. It maintains favorable stability under low signal-to-noise ratio working conditions, and can accurately identify single and compound faults of machine tool bearings, providing technical support for fault early warning, operation and maintenance guarantee of machine tool equipment.
Downloads
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
S. Wang, Z. Yu, F. Huo, et al., “The intelligent operation and maintenance method and experimental research of CNC machine tool bearings,” Proc. Inst. Mech. Eng. Part E J. Process Mech. Eng., vol. 240, no. 3, pp. 2540– 2553, 2026, DOI: 10.1177/09544089241284483.
W. Zhou, J. Zhou, C. Li, et al., “Research on compound fault diagnosis of rolling bearing of CNC machine tools based on parameter optimization MOMEDA,” Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci., vol. 239, no. 13, pp. 5284–5296, 2025, DOI: 10.1177/09544062251324346.
Y. Li, X. Liang, L. Na, et al., “DARTS-CNN-BiLSTM: Intelligent Fault Diagnosis for Computer Numerical Control Machine Tool Feed System,” Ann. N. Y. Acad. Sci., vol. 1560, no. 1, p. e70317, 2026, DOI: 10.1111/NYAS.70317.
M. Y. Roopa, S. K. Latha, B. B. Reddy, et al., “DiGACN: Distributed Convolutional Network for Anomaly Detection in Computer Numerical Control Machines,” Iran. J. Sci. Technol. Trans. Electr. Eng., pp. 1–31, 2026, DOI: 10.1007/S40998-026-01144-W.
L. Junying, “Fault Detection Method of CNC Machine Tool,” Parallel Process. Lett., vol. 32, no. 2, 2022, DOI: 10.1142/S0129626421410012.
M. Kumbhar, S. Bandaru, A. Karlsson, “Condition Monitoring of a Machine Tool Ballscrew Using Wavelet Transform based Unsupervised Learning,” Procedia CIRP, vol. 130, pp. 342–347, 2024, DOI: 10.1016/J.PROCIR.2024.10.098.
Z. Ni, Y. Tong, Y. Song, et al., “Enhanced Bearing Fault Diagnosis in NC Machine Tools Using Dual-Stream CNN with Vibration Signal Analysis,” Processes, vol. 12, no. 9, p. 1951, 2024, DOI: 10.3390/PR12091951.
J. Sun, Z. Liu, C. Qiu, et al., “An axial attention-BiLSTM-based method for predicting the migration of CNC machine tool spindle thermal error under varying working conditions,” Int. J. Adv. Manuf. Technol., vol. 130, nos. 3–4, pp. 1405–1419, 2024, DOI: 10.1007/S00170-023-12759-2.
X. Yang, S. Peng, Z. Zhang, et al., “Thermal error prediction in dry hobbing machine tools: A CNN-BiGRU network with spatiotemporal feature fusion,” Measurement, vol. 256, p. 118389, 2025, DOI: 10.1016/J.MEASUREMENT.2025.118389.
Z. Xiaochen, W. Di, X. Zhuofan, et al., “Comparative Study on Surface Contact Fatigue of Machine Tool Bearing Raceways with Different Rolling Elements: Two Failure Bearings Case Studies,” Tribol. Lett., vol. 71, no. 2, 2023, DOI: 10.1007/S11249-023-01722-7.
X. Gu, Y. Tian, C. Li, et al., “Improved SE-ResNet Acoustic–Vibration Fusion for Rolling Bearing Composite Fault Diagnosis,” Appl. Sci., vol. 14, no. 5, 2024, DOI: 10.3390/app14052182.
S. Zhang, B. Song, Z. Fupeng, et al., “A decoupling method for compound faults of rolling bearings based on multi-modal bidirectional cross-attention fusion and contrastive learning,” IOP Publ., 2026, DOI: 10.1088/1361-6501/ae080f.
T. Nabil, Z. Zoubir, Y. Ramdane, et al., “Intelligent fault classification in wind turbine bearings using multi-resolution wavelet analysis and artificial neural networks,” Results Eng., vol. 30, p. 108661, 2026, DOI: 10.1016/J.RINENG.2025.108661.
X. Yang, R. Hou, C. Liu, et al., “A Novel Rolling Bearing Fault Diagnosis Approach Integrating Improved Wavelet Threshold Denoising and CEHFHA-EMD,” J. Vib. Eng. Technol., vol. 14, no. 2, p. 69, 2026, DOI: 10.1007/S42417-025-02185-X.
D. Yadav, D. Rani, P. O. Verma, “Hybridization of attention mechanism based CNN bi-directional LSTM model for enhancing HAR,” Signal Image Video Process., vol. 19, no. 12, p. 1017, 2025, DOI: 10.1007/S11760-025-04520-X.
Y. Zheng, G. Fu, S. Mu, et al., “Thermal Error Transfer Prediction Modeling of Machine Tool Spindle with Self-Attention Mechanism-Based Feature Fusion,” Machines, vol. 12, no. 10, p. 728, 2024, DOI: 10.3390/MACHINES12100728.