2025 Vol.8 Oct.N05

• Title: Knowledge Graph-Constrained Neural Differential Equation Modeling for Gearbox Fault Diagnosis
• Name: Haiyong Cao,Cuicui Dong
• Company: School of Computer Information Engineering, Nanchang Institute of Technology,Nanchang,330044,China
• Abstract:

  Gearbox fault diagnosis methods predominantly rely on discrete differential equations or statistical signal processing, achieving fault identification by extracting time-domain features such as vibration and temperature to construct classification models. While these approaches capture steady-state characteristics, their static modeling frameworks struggle to depict the dynamic evolution of faults, resulting in suboptimal diagnostic accuracy. To address this, we propose knowledge graph-constrained neural differential equation fault modeling for gearboxes. Through domain expert consultation and data mining, entities such as gears and bearings are defined, forming a triplet set that constitutes the topological structure of the knowledge graph. Embedding models like TransE map entities and relationships to low-dimensional vector spaces while preserving semantic associations. A framework of ordinary differential equations is constructed using gearbox state variables. Subsequently, fault propagation rules are extracted via knowledge graph embedding techniques and converted into dynamic constraint terms, enabling dynamic characterization of gearbox nonlinearity. By introducing latent variables to encode structured constraints from the knowledge graph and combining these with knowledge graph constraints, fault probabilities are computed, thereby achieving gearbox fault diagnosis. Experimental validation confirms the proposed method's diagnostic accuracy. Comparative test results demonstrate that when applying this approach to gearbox fault modeling and diagnosis via neural differential equations, the algorithm achieves a fault label overlap rate of 92.3%, delivering highly satisfactory diagnostic performance.

  
  

• Keyword: knowledge graph; gearbox; neural differential equation; fault modeling; fault diagnosis;
• DOI: 10.12250/jpciams2025091016
• Citation form: Haiyong Cao,Cuicui Dong.Knowledge Graph-Constrained Neural Differential Equation Modeling for Gearbox Fault Diagnosis[J]. Computer Informatization and Mechanical System,2025,Vol.8,pp.

[1] Luo D, Yuan W. Gearbox Fault Diagnosis of Wind Turbines Based on Attention and Multi-scale CNN. Proceedings of the 2024 8th International Conference on Electronic Information Technology and Computer Engineering, 2024:1385-1391.

[2] Huang W, Wang J, Du G, et al. Balance Sparse Decomposition Method with Nonconvex Regularization for Gearbox Fault Diagnosis[J]. Chinese Journal of Mechanical Engineering, 2024, 37(1):1-14.

[3] Wang Y. Gearbox Fault Diagnosis Based on ResNet50-MSVEA-GATv2 Model[J]. 2024 8th International Conference on Electrical, Mechanical and Computer Engineering (ICEMCE), 2024:1318-1324.

[4] Zhang B, Miao Y, Lin J, et al. Application of CYCBD for the planetary gearbox fault diagnosis based on built-in encoder signal[J]. Equipment Intelligent Operation and Maintenance, 2025:389-397.

[5] Xie K, Chen L. A Wind Gearbox Fault Detection Method Based on a Lightweight Attention Network[J]. 2024 Global Reliability and Prognostics and Health Management Conference (PHM-Beijing), 2024:1-5.

[6] Cao X, Li Y, Yu K, et al. Multimodal industrial gearbox fault diagnosis based on vibration signals and infrared thermal imaging[J]. Proceedings of the 2024 2nd International Conference on Frontiers of Intelligent Manufacturing and Automation, 2024:193-196.  

[7] Guo J, He Q, Ball GAD. A local modulation signal bispectrum for multiple amplitude and frequency modulation demodulation in gearbox fault diagnosis[J]. Structural Health Monitoring, 2023, 22(5):3189-3205.

[8] Huang P, Gu Y, Qiu G. A novel feature dimensionality reduction method for gearbox fault diagnosis with HMSDE, DANCo-DDMA and KELM[J]. Nonlinear Dynamics, 2024, 112(16):14071-14091.

[9] Lou P, Zhang D, Yin K, et al. Sparsity signal decomposition via GMC penalty for aero-engine accessory gearbox fault diagnosis[J]. Equipment Intelligent Operation and Maintenance, 2025:583-591.

[10] Wang H, Liu Z, Shi W L. A Gearbox Fault Diagnosis Method Based on Graph Neural Networks and Markov Transform Fields[J]. IEEE Sensors Journal, 2024, 24(15):25186-25196.

[11] Li M, Wang Y, Wang J, et al. Varying-Parameter Time-Frequency Manifold Signal Reconstruction for Gearbox Fault Diagnosis[J]. 2024 Global Reliability and Prognostics and Health Management Conference (PHM-Beijing), 2024:1-5. 

[12] Suryawanshi S, Ghorpade U, Bhosale D G, et al. Application of Machine Learning and Deep Learning in Gearbox Fault Diagnosis: A Comprehensive Review[J]. Lecture Notes in Networks and Systems, 2025:343-353.

[13] Saleem O, Manzoor A. Fatigue Failure in a Pinion Shaft of the Planetary Gearbox[J]. Journal of Failure Analysis and Prevention, 2025, 25(1):47-61.

[14] Liu X, Zhang X, Suo Y, et al. Fault Diagnosis of Small Sample Wind Turbine Gearbox Based On T-CBC[J]. 2024 10th International Conference on Systems and Informatics (ICSAI), 2024:1-6.

[15] Hemati A, Shooshtari A. Root Cause Failure Analysis of Planetary Gearbox Vibration Using Velocity Spectra and Time Signal Analysis[J]. Journal of Failure Analysis and Prevention, 2025, 25(1):502-517.