2026 Vol.9 Apr.N02

• Title: Fault Diagnosis of Oil-immersed Transformer Based on MPC-optimized Bayesian Network
• Name: Hang Qi1, Zhengrui Qiao1, Jingzhao Chen1, 2, Jin Niu1
• Company: (1. Faculty of Engineering, SIAS University, Zhengzhou Henan, 451150, China) (2. School of Artificial Intelligence and Automation, Wuhan University of Science and Technology, Wuhan Hubei, 430081, China)
• Abstract:

  Oil-immersed transformers are the core of power systems, and their fault diagnosis accuracy is critical to system operation. Aiming at the deficiencies of the traditional three-ratio method in DGA and the insufficient analysis of multi-state quantity correlations by existing machine learning methods, this paper proposes a fault diagnosis method based on Bayesian network optimized by MPC algorithm. Based on DGA, 5 characteristic gases are selected to extract 9-dimensional fault features via the non-coded ratio method with normalization, and a Bayesian network model is constructed. The MPC algorithm optimizes the network structure by adding cyclic graph detection after orientation rule execution, avoiding cyclic graph generation. Experimental verification is conducted on DGA data with 80% for training and 20% for testing, and comparisons are made with SVM, XGBoost and KNN. The results show that the proposed model attains high diagnostic accuracy for individual samples, outperforms comparative models in diagnosing all fault types, and significantly enhances the accuracy and reliability of transformer fault diagnosis, providing an effective new approach for transformer condition monitoring.

  
  

• Keyword: Bayesian network; modified principal component analysis algorithm; dissolved gas analysis in oil; the transformer fault diagnosis
• DOI: 10.12250/jpciams2026090403
• Citation form: Hang Qi, Zhengrui Qiao, Jingzhao Chen, Jin Niu.Fault Diagnosis of Oil-immersed Transformer Based on MPC-optimized Bayesian Network[J]. Computer Informatization and Mechanical System,2026,Vol.9,pp.

[1] Wang F, Wang X W, Ke Y, et al. Transformer fault diagnosis method based on improved EPO-BP neural network[J]. Northeast Electric Power Technology, 2026, 47(01):43-48.

[2] Yuan F T, Jian S K, Jiang Y Q, et al. Fault diagnosis method for oil-immersed transformer windings based on multi-physics and improved convolutional neural network[J]. Smart Power, 2026, 54(02):106-114.

[3] Xu B Q, Wang B, Sun L L, et al. Fault early warning of wind turbine based on improved CEEMD and enhanced vision Transformer model[J]. Transactions of China Electrotechnical Society, 2025, 40(20):6537-6551.

[4] Zhang Z W, Yao J, Zhao W, et al. Fault diagnosis method of mining transformer based on improved wavelet threshold and CNN[J]. Electric Engineering, 2025(21):261-264+267.

[5] Ma Y H, Wang X C, Li G J, et al. Research on fault diagnosis method of traction transformer based on ISSA optimized SVM[J]. Journal of the China Railway Society, 2026, 48(02):48-55.

[6] Xiao J H, Dong Y H, Huang H, et al. Prediction model of hot spot temperature in transformer windings based on artificial neural network[J]. Information Technology, 2025(08):177-183+189.

[7] ZHENG Y B, SUN C X, LI J, et al. Association rule analysis of transformer fault characteristic quantity reliability[J]. High Volt Technol. 2012,38,87-93.

[8] Zhang Y J, Ma J S, Wei Y, et al. Research on thrust prediction method of aero-engine based on Bayesian optimized neural network[J]. Journal of Propulsion Technology, 2026:1-16.

[9] Wilhelm Y, Reimann P, Gauchel W, et al. An ontology-driven bayesian network approach to fault diagnosis and correction in manufacturing[J]. The International Journal of Advanced Manufacturing Technology, 2026: 1-24.

[10] Gu M H, Li Y L, Wang Y J, et al. Research of fault diagnosis and prediction based on Bayesian networks and time series model for meta-action[J]. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 2026, 240(1-2): 282-295.

[11] Dai T, Li X H, Sui Y, et al. A new dynamic Bayesian network model for fault diagnosis of complex systems with high uncertainty in nuclear power plant[J]. Energy, 2026, 342: 139548.

[12] MICHAIL T. Bayesian network learning with the PC algorithm: an improved and correct variation[J]. Applied Artificial Intelligence. 2018,1087-6545.