Bridging the Accuracy–Robustness Gap in AI-Based Dissolved Gas Analysis for Power Transformer Diagnostics: A Critical Review
Source abstract
Abstract
Dissolved gas analysis (DGA) remains a cornerstone technique for the early detection of internal faults in power transformers. However, the interpretation of gas signatures is inherently complex due to coupled thermochemical processes, operating variability, and measurement uncertainties. Over the past three decades, artificial intelligence (AI) techniques have been widely applied to DGA-based diagnostics, evolving from conventional machine learning to advanced deep learning and hybrid models. Although many studies report diagnostic accuracies exceeding 95%, these results often rely on limited datasets and heterogeneous validation practices, raising concerns about their robustness in real-world operating conditions. This review provides a comprehensive and critical analysis of the evolution of AI-based approaches, with emphasis on methodological foundations rather than algorithmic performance alone. Key aspects examined include dataset construction, validation protocols, class imbalance handling, and experimental design. The analysis highlights a structural Accuracy-Robustness Gap, in which high reported accuracies often coexist with limited evidence of generalisation under realistic industrial variability. To address this issue, a conceptual framework is proposed linking the physical complexity of the oil–paper insulation system, dataset characteristics, and algorithmic modeling strategies. The review further identifies several key research gaps in benchmarking practices, dataset availability, robustness evaluation, and model interpretability. Based on these findings, a standardized evaluation framework is proposed to support more rigorous and comparable assessments of AI-based DGA diagnostic models. Future research directions are discussed, highlighting the need for physics-informed learning, distribution-aware validation, and shared datasets to enable reliable deployment of AI-based transformer diagnostics.
Source: crossref · Quality: baseline
Affiliations
Canada Research Chair Tier 1, in Aging of Oil-Filled Equipment on High Voltage Lines (ViAHT), University of Quebec at Chicoutimi, Saguenay, QC G7H 2B1, Canada
Hydro-Québec (HQ), Varennes, QC J3X 1S1, Canada
Rio Tinto, Saguenay, QC G7S 2H8, Canada
Funding
Canada Research Chair Tier 1 in Aging of Oil-Filled Equipment on High Voltage Lines · Natural Sciences and Engineering Research Council of Canada
References (100+)
Show source-provided references
1. Conventional methods of dissolved gas analysis using oil-immersed power transformer for fault diagnosis: A review10.1016/j.epsr.2022.109064 ↗
2. (2019). IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers (Standard No. C57.104-2019).
3. (2022). Mineral Oil-Filled Electrical Equipment in Service—Guidance on the Interpretation of Dissolved and Free Gases Analysis (Standard No. IEC 60599:2022).
4. Lai, J., Weng, D., Xian, F., Xie, Y., Chen, Y., Zhou, Q., and Yuan, C. (2026). A Hybrid Optimization Model for Transformer Fault Diagnosis Based on Gas Classification. Digital, 6.10.3390/digital6010024 ↗
5. Pt-modified Janus MoSeTe monolayer: A first-principles study for transformer oil dissolved gas sensing10.1016/j.colsurfa.2026.140259 ↗
6. Transformer fault diagnosis using machine learning: A method combining SHAP feature selection and intelligent optimization of LGBM10.1186/s42162-025-00519-3 ↗
7. Influence of data balancing on transformer DGA fault classification with machine learning algorithms10.1109/TDEI.2022.3230377 ↗
8. Traditional fault diagnosis methods for mineral oil-immersed power transformer based on dissolved gas analysis: Past, present and future10.1049/nde2.12082 ↗
9. Improving the reliability of dissolved gas analysis (DGA) diagnostics through consideration of measurement device uncertainty10.1109/TDEI.2024.3495585 ↗
10. Fault diagnosis of power transformer based on multi-layer SVM classifier10.1016/j.epsr.2004.07.013 ↗
11. Romadhona, G., Ismara, K.I., and Nugraheni, M. (2026). Dissolved Gas Analysis in Vegetable Oil-Insulated Transformers: A Systematic Literature Review of Methods and Effectiveness. IEEE Trans. Dielectr. Electr. Insul., 1.
12. Min, L.H., and Chang, C. (2009). Application of Dempster-Shafer’s theory of evidence for transformer incipient fault diagnosis. Proceedings of the 8th International Conference on Advances in Power System Control, Operation and Management (APSCOM 2009), Hong Kong, China, 8–11 November 2009, Institution of Engineering and Technology.
13. A Hybrid Ensemble Approach for Dissolved Gas Analysis-Based Power Transformer Fault Diagnosis Using SMOTE and GBDT10.1007/s13369-025-10772-z ↗
14. Meira, M., Álvarez, R., Verucchi, C., and Catalano, L. (2026). Thermal Fault Model Validated in Mineral Oil for Dissolved Gas Analysis. IEEE Trans. Dielectr. Electr. Insul.10.1109/TDEI.2026.3658324 ↗
15. Investigate transformer fault diagnosis performance of dissolved gas analysis with measurement error10.1080/15325008.2017.1310955 ↗
16. Combining and comparing various machine-learning algorithms to improve dissolved gas analysis interpretation10.1049/iet-gtd.2018.0059 ↗
17. A CatBoost-Based Framework for Data-Driven Fault Diagnosis in Power Transformers Using Dissolved Gas Analysis10.1016/j.rineng.2026.109972 ↗
18. Patil, A.J., Malik, H., Singh, A., Gupta, P.P., Naresh, R., Jarial, R.K., Patil, M.A., and Kalkhambkar, V. (2026). Enhancing Severity Level Detection of Transformer Discharge and Thermal Faults Using Artificial Intelligence in Duval Triangle 1. IEEE Trans. Dielectr. Electr. Insul.10.1109/TDEI.2025.3650517 ↗
19. Enhancing transformer health monitoring with ai-driven prognostic diagnosis trends: Overcoming traditional methodology’s computational limitations10.1109/MIE.2023.3329277 ↗
20. Transformer-fault diagnosis by integrating field data and standard codes with training enhancible adaptive probabilistic network10.1049/ip-gtd:20040833 ↗
21. Artificial immune network classification algorithm for fault diagnosis of power transformer10.1109/TPWRD.2007.893182 ↗
22. A probabilistic classifier for transformer dissolved gas analysis with a particle swarm optimizer10.1109/TPWRD.2008.915812 ↗
23. Application of SVM and KNN to Duval Pentagon 1 for transformer oil diagnosis10.1109/TDEI.2017.006841 ↗
24. Transformer fault diagnosis based on multi-class AdaBoost algorithm10.1109/ACCESS.2021.3135467 ↗
25. A fault diagnostic method for oil-immersed transformer based on multiple probabilistic output algorithms and improved DS evidence theory10.1016/j.ijepes.2021.107828 ↗
26. Baker, E., Nese, S.V., and Dursun, E. (2023). Hybrid condition monitoring system for power transformer fault diagnosis. Energies, 16.10.3390/en16031151 ↗
27. Transformer fault diagnosis in modern power systems using a genetic algorithm-optimized neural network and dissolved gas analysis10.1007/s43926-026-00296-9 ↗
28. Bouzar, M., Fofana, I., and Picher, P. (October, January 29). A Novel Hybrid Multilayer Model for Diagnosing Faults in Power Transformers Through Dissolved Gas Analysis (DGA). Proceedings of the CIGRÉ 2025 International Symposium, Montréal, QC, Canada.
29. A transformer fault diagnosis method based on hybrid improved grey wolf optimization and least squares-support vector machine10.1049/gtd2.12405 ↗
30. A transformer fault diagnosis method based on parameters optimization of hybrid kernel extreme learning machine10.1109/ACCESS.2021.3112478 ↗
31. Feature selection by joint manifold structures for transformer fault diagnostic10.1016/j.epsr.2026.112906 ↗
32. Twin Network Approach for Multi-Fault Classification in Power Transformers Using Enriched Features and Zero Synthetic Data10.1109/TDEI.2025.3563316 ↗
33. Power transformer fault diagnosis under measurement originated uncertainties10.1109/TDEI.2012.6396956 ↗
34. Improvement of power transformer insulation diagnosis using oil characteristics data preprocessed by SMOTEBoost technique10.1109/TDEI.2014.004547 ↗
35. Dissolved gas analysis method based on novel feature prioritisation and support vector machine10.1049/iet-epa.2014.0085 ↗
36. A semi-supervised fault diagnosis method for transformers based on discriminative feature enhancement and adaptive weight adjustment10.1109/TIM.2024.3398079 ↗
37. Zeng, B., Guo, J., Zhu, W., Xiao, Z., Yuan, F., and Huang, S. (2019). A transformer fault diagnosis model based on hybrid grey wolf optimizer and LS-SVM. Energies, 12.10.3390/en12214170 ↗
38. Research on transformer fault diagnosis method based on ACGAN and CGWO-LSSVM10.1038/s41598-024-68141-z ↗
39. Transformer fault diagnosis using continuous sparse autoencoder10.1186/s40064-016-2107-7 ↗
40. Dissolved gas analysis of insulating oil for power transformer fault diagnosis with deep belief network10.1109/TDEI.2017.006727 ↗
41. Power transformer fault diagnosis based on DGA using a convolutional neural network with noise in measurements10.1109/ACCESS.2021.3102415 ↗
42. Wu, X., He, Y., and Duan, J. (2020). A deep parallel diagnostic method for transformer dissolved gas analysis. Appl. Sci., 10.10.3390/app10041329 ↗
43. Fault diagnosis of power transformers using graph convolutional network
44. Sensing incipient faults in power transformers using bi-directional long short-term memory network10.1109/LSENS.2022.3233135 ↗
45. El-kenawy, E.-S.M., Albalawi, F., Ward, S.A., Ghoneim, S.S., Eid, M.M., Abdelhamid, A.A., Bailek, N., and Ibrahim, A. (2022). Feature selection and classification of transformer faults based on novel meta-heuristic algorithm. Mathematics, 10.10.3390/math10173144 ↗
46. Yazdani, M.H., Azad, M.M., Jo, S.-H., and Kim, H.S. (2026). Advancements in Computational Methods for Intelligent Prognostics and Health Management of Power Transformers. Arch. Comput. Methods Eng., 1–37.10.1007/s11831-026-10548-0 ↗
47. Incipient fault diagnosis in power transformers by data-driven models with over-sampled dataset10.1016/j.epsr.2021.107519 ↗
48. Identifying Faults in Power Transformers Based on Machine-Learning Algorithms Compared With Other Techniques10.1002/ese3.70521 ↗
49. A Power Transformer Fault Diagnosis Method Based on Improved Variational Quantum Shadow Learning10.1109/TPWRD.2025.3546722 ↗
50. Few-Shot power transformers fault diagnosis based on Gaussian prototype network10.1016/j.ijepes.2024.110146 ↗
51. Transformer Fault Diagnosis Method Based on Pseudo-Image Processing and Improved Random Forest10.1002/tee.70164 ↗
52. Deep machine learning-based asset management approach for oil-immersed power transformers using dissolved gas analysis10.1109/ACCESS.2024.3366905 ↗
53. Review on the research and practice of deep learning and reinforcement learning in smart grids10.17775/CSEEJPES.2018.00520 ↗
54. From limited data to reliable diagnosis: An interpretable deep learning framework for transformer fault analysis10.1016/j.ijepes.2025.111227 ↗
55. Shang, H., Wei, Y., and Zhang, S. (2025). Transformer Fault Diagnosis Based on Knowledge Distillation and Residual Convolutional Neural Networks. Entropy, 27.10.3390/e27070669 ↗
56. Hierarchical federated learning for power transformer fault diagnosis10.1109/TIM.2022.3196736 ↗
57. PS-DFL: A Parallel Sequential Decentralized Federated Learning Framework to Facilitate DGA Interpretation Related to Transformer Fault Diagnosis10.1109/TDEI.2025.3568441 ↗
58. Deep learning for fault diagnosis in power transmission lines: Current trends, limitations, and future directions10.1109/ACCESS.2025.3628908 ↗
59. Privacy-preserving federated learning for power transformer fault diagnosis with unbalanced data10.1109/TII.2023.3333914 ↗
60. Precise transformer fault diagnosis via random forest model enhanced by synthetic minority over-sampling technique10.1016/j.epsr.2023.109361 ↗
61. Bayesian duval triangle method for fault prediction and assessment of oil immersed transformers10.1177/0020294017707461 ↗
62. Application of Parzen Window estimation for incipient fault diagnosis in power transformers10.1049/hve.2018.5061 ↗
63. Accurate sensing of power transformer faults from dissolved gas data using random forest classifier aided by data clustering method10.1109/JSEN.2022.3149409 ↗
64. Novel prediction-reliability based graphical DGA technique using multi-layer perceptron network & gas ratio combination algorithm10.1049/iet-smt.2018.5397 ↗
65. Machine learning based multi-method interpretation to enhance dissolved gas analysis for power transformer fault diagnosis10.1016/j.heliyon.2024.e25975 ↗
66. Detecting Incipient Faults in Power Transformers through Hybrid Model of DGA and Machine Learning10.1016/j.epsr.2026.112877 ↗
67. A robust explainable machine learning pipeline for transformer health index prediction addressing data pathologies and redundancy10.1016/j.epsr.2026.113275 ↗
68. Data-driven Fault Detection and Diagnosis of Power Transformers Using Dissolved Gas Analysis and Random Forest Classifiers
69. Gaussian process multi-class classification for transformer fault diagnosis using dissolved gas analysis10.1109/TDEI.2021.009470 ↗
70. A novel power transformer fault diagnosis model based on harris-hawks-optimization algorithm optimized kernel extreme learning machine10.1007/s42835-022-01000-x ↗
71. Fault diagnosis of oil-immersed power transformer based on difference-mutation brain storm optimized catboost model10.1109/ACCESS.2021.3135283 ↗
72. An interpretable machine learning method for fault diagnosis of oil-immersed transformers based on edge inference10.1016/j.ijepes.2025.110647 ↗
73. Explainable Fault Diagnosis of Oil-Immersed Transformers: A Glass-Box Model10.1109/TIM.2024.3350131 ↗
74. Ji, N., Chen, X., Qin, X., Wei, W., Jiang, C., Bo, Y., and Tao, K. (2024). Transformer fault identification based on GWO-optimized Dual-channel M-A method. PLoS ONE, 19.10.1371/journal.pone.0312474 ↗
75. A Novel Transformer Fault Diagnosis Method for Imbalanced Data Based on Dual Attention Mechanism and IEC Ratio Features10.1109/TDEI.2025.3576802 ↗
76. Transformer fault diagnosis method based on MTF and GhostNet10.1016/j.measurement.2025.117056 ↗
77. DGA Fault Diagnosis Method Based on Gaussian Copula Data Augmentation and Transfer Learning10.1007/s42835-025-02409-w ↗
78. Zhang, X., Yang, K., and Zheng, L. (2024). Transformer Fault Diagnosis Method Based on TimesNet and Informer. Actuators, 13.10.3390/act13020074 ↗
79. Artificial intelligent fault diagnostic method for power transformers using a new classification system of faults10.1007/s42835-019-00105-0 ↗
80. Detection of carbonization of paper in transformers using duval pentagon 2 and triangle 510.1109/TDEI.2023.3278623 ↗
81. Identification of arcing faults in paper and oil in transformers—Part I: Using the Duval pentagons10.1109/MEI.2022.9648268 ↗
82. Evolution of graphical methods for the identification of insulation faults in oil-immersed power transformers: A review10.1016/j.rser.2024.114473 ↗
83. Ibrahim, R.A., and Hebala, A. (2025). A Feature-Enhanced Approach to Dissolved Gas Analysis for Power Transformer Health Prediction Through Interpretable Ensemble Learning and Multi-Model Evaluation. Technologies, 14.10.3390/technologies14010006 ↗
84. A dissolved Gases Analysis Method for Power Transformer Faults Diagnosis Based on the Observation of Subsets of Labelled Fault Data10.1007/s42835-025-02144-2 ↗
85. Analyzing the Impact of TiO2 Nanoparticles in Transformer Oil-Based Nanofluids Through Dissolved Gas Analysis10.1109/ACCESS.2026.3685071 ↗
86. Duval, C.M. (2019). Advances in DGA Interpretation. CIGRE Technical Brochure #771, CIGRE. CIGRE D1-01/A2.47.
87. Cao, H., Zhou, C., Meng, Y., Shen, J., and Xie, X. (2024). Advancement in transformer fault diagnosis technology. Front. Energy Res., 12.10.3389/fenrg.2024.1437614 ↗
88. Dukarm, J.J. (1993). Transformer oil diagnosis using fuzzy logic and neural networks. Canadian Conference on Electrical and Computer Engineering (CCECE), Institute of Electrical and Electronics Engineers.10.1109/CCECE.1993.332323 ↗
89. A fuzzy diagnostic system for incipient transformer faults based on DGA of the insulating transformer oils10.15866/iree.v11i3.8453 ↗
90. Intelligent Transformer Fault Identification Using GALO-FIS and Refined Ratio Techniques10.1109/TPWRD.2025.3612218 ↗
91. A new model of faults classification in power transformers based on data optimization method10.1016/j.epsr.2021.107446 ↗
92. Interpretation of gas-in-oil analysis using new IEC publication 60599 and IEC TC 10 databases10.1109/57.917529 ↗
93. A Transformer Fault Diagnosis Method Based on Data Augmentation and IEVO-CNN10.1109/TIM.2025.3566811 ↗
94. A Method of Kernel Principal Component Analysis and Machine Learning Algorithms for Fault Diagnosis of Power Transformers10.1109/TDEI.2025.3543337 ↗
95. Wang, J., Chi, J., Ding, Y., Yao, H., and Guo, Q. (2025). Based on PCA and SSA-LightGBM oil-immersed transformer fault diagnosis method. PLoS ONE, 20.10.1371/journal.pone.0337958 ↗
96. Fault diagnosis method for transformer based on NCA and CapSA-RELM10.1007/s00202-023-01974-5 ↗
97. A novel double-stacked autoencoder for power transformers DGA signals with an imbalanced data structure10.1109/TIE.2021.3059543 ↗
98. Classification of fault and stray gassing in transformer by using duval pentagon and machine learning algorithms10.1007/s13369-022-06770-0 ↗
99. The improvement of SMOTE-ENN-XGBoost through Yeo Johnson strategy on Dissolved Gas Analysis dataset10.1016/j.egyr.2025.05.013 ↗
100. Lv, Y., Huang, H., Wang, Y., and Zhang, D. (2026). Adaptive Semi-Supervised Random Forest for Transformer Fault Diagnosis in the case of limited labeled data. IEEE Trans. Dielectr. Electr. Insul.10.1109/TDEI.2026.3687172 ↗