Article Details

Abstract

To address the problems of relying solely on vibration data and insufficient interpretability of results in rotating machinery fault diagnosis, this study proposes a diagnostic method integrating the Multi-scale Residual Attention Convolutional Neural Network (MRTCNN) with knowledge graph. Given raw bearing vibration data and mechanism knowledge as inputs, the method first conducts entity extraction and label normalization; subsequently, an end-to-end MRTCNN is constructed to complete feature representation and fault identification; finally, the diagnostic results, working conditions, and mechanism events are written into a graph database to form a traceable bearing fault knowledge graph, realizing fine-grained information display and providing auxiliary decision support. Validation on two datasets with a unified data processing workflow shows that, for tasks involving up to 160 fault types, the weighted F1-score of MRTCNN is improved by 12.07% on average compared with benchmark models. Additionally, the proposed method demonstrates higher stability and cross-working-condition generalization ability in conventional fault diagnosis experiments and multi-algorithm comparisons.

Keywords

knowledge graph; multi-scale convolution; fault diagnosis; bearing

References

1. [1] Cui K，Liu M，Meng Y. A New Fault Diagnosis of Rolling Bearing on FFT Image Coding and L-CNN [J/OL].Measurement Science and Technology，2024，35（7）：076108.https://iopscience.iop.org/article/10.1088/1361-6501/ad3295.
2. [2] Li X，Jiao R，Zhou C，et al. Multi-Level Contrastive Reasoning over Knowledge Graphs for Fault Diagnosis of Confounding Cases [J/OL].Neurocomputing，2025：130990.https://doi.org/10.1016/j.neucom.2025.130990.
3. [3] Li X，Zhang W，Ding Q，et al. Diagnosing Rotating Machines with Weakly Supervised Data Using Deep Transfer Learning [J].IEEE Transactions on Industrial Informatics，2020，16（3）：1688-1697.
4. [4] Zhang Shuntang, Xu Jiayu, Song Haicao Research on Intelligent Diagnosis of Mechanical Arm Faults Based on Deep Learning Networks and Domain Knowledge Graph [J]. Machine Tool and Hydraulic, 2026, 54 (3): 85-94
5. [5] Mo C，Han H，Liu M，et al. Research on SVM-Based Bearing Fault Diagnosis Modeling and Multiple Swarm Genetic Algorithm Parameter Identification Method [J/OL].Mathematics，2023，11（13）：2864.https://doi.org/10.3390/math11132864.
6. [6] Singh M T. Graph-Based Fault Diagnosis for Rotating Machinery: Adaptive Segmentation and Structural Feature Integration [J/OL].Results in Engineering，2025，27：106566.https://doi.org/10.1016/j.rineng.2025.106566.
7. [7] Tian X，Gan H，Liu Y. Construction of Knowledge Graph for Marine Diesel Engine Faults Based on Deep Learning Methods [J/OL].Journal of Marine Science and Engineering，2025，13（4）：693.https://doi.org/10.3390/jmse13040693.
8. [8] Chen H，Wang X B，Yang Z X. Knowledge Graph for Fault Diagnosis of Mechanical System [C]//2022 IEEE 11th Data Driven Control and Learning Systems Conference (DDCLS)，2022：795-799.
9. [9] Kuncan M，Kaplan K，Naz M，et al. A Novel Feature Extraction Method for Bearing Fault Classification with One Dimensional Ternary Patterns [J/OL].ISA Transactions，2019，100.https://doi.org/10.1016/j.isatra.2019.11.006.
10. [10] Schölkopf B，Smola A J. Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond [M].MIT Press，2001.
11. [11] Liu Y，Zhou Y，Liu Y，et al. Intelligent Fault Diagnosis for CNC through the Integration of Large Language Models and Domain Knowledge Graphs [J].Engineering，2025，52：311-322.
12. [12] Li Zhibo, Li Yuanyuan, Cai Yin Combination of Convolutional Neural Networks and Knowledge Graph for Bearing Fault Diagnosis [J]. Noise and Vibration Control, 2024, 44 (2): 156-163
13. [13] Usama M，Ahmad B，Song E，et al. Attention-Based Sentiment Analysis Using Convolutional and Recurrent Neural Network [J].Future Generation Computer Systems，2020，113：571-578.
14. [14] Yang Y，Zheng H，Li Y，et al. A Fault Diagnosis Scheme for Rotating Machinery Using Hierarchical Symbolic Analysis and Convolutional Neural Network [J].ISA Transactions，2019，91：235-252.
15. [15] Xu Y，Li Z，Wang S，et al. A Hybrid Deep-Learning Model for Fault Diagnosis of Rolling Bearings [J/OL].Measurement，2021，169（6）：108502.https://doi.org/10.1016/j.measurement.2020.108502.
16. [16] Liang P，Deng C，Wu J，et al. Compound Fault Diagnosis of Gearboxes via Multi-Label Convolutional Neural Network and Wavelet Transform [J/OL].Computers in Industry，2019，113（2）：103132.https://doi.org/10.1016/j.compind.2019.103132.
17. [17] Szegedy C，Liu W，Jia Y，et al. Going Deeper with Convolutions [C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)，2014：1-9.
18. [18] Zhao Xiaoqiang, Zhang Yazhou A fault diagnosis method for rolling bearings under variable operating conditions using improved convolutional neural networks [J]. Journal of Xi'an Jiaotong University, 2021, 55 (12): 108-118
19. [19] He K，Zhang X，Ren S，et al. Deep Residual Learning for Image Recognition [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition， 2016：770-778.
20. [20] Liu Junpeng, Huang Kaiyu, Li Jiuyi, etc Neural Machine Translation Based on Multi Coverage Model [J]. Journal of Software, 2022, 33 (3): 1141-1152

Funding

Open Fund of Guangdong Key Laboratory of Modern Control Technology (GDKLMCT202401); National Natural Science Foundation of China (62073091)

About the authors

Luo Pengfei (2002-), male, Han ethnicity, from Hengyang, Hunan, master's student, research interests include bearing fault diagnosis; Liu Mei (1967-), female, Han ethnicity, from Lianjiang, Guangdong, professor, master's supervisor, doctoral degree, research interests include artificial intelligence and its applications.