Uncertainty-aware fusion of probabilistic classifiers for improved transformer diagnostics

Aizpurua, Jose Ignacio and Catterson, Victoria M. and Stewart, Brian G. and McArthur, Stephen D. J. and Lambert, Brandon and Cross, James G. (2021) Uncertainty-aware fusion of probabilistic classifiers for improved transformer diagnostics. IEEE Transactions on Systems, Man and, Cybernetics: Systems, 51 (1). pp. 621-633. ISSN 2168-2216 (https://doi.org/10.1109/TSMC.2018.2880930)

[thumbnail of Aizpurua-etal-TSMCS2018-Uncertainty-aware-fusion-of-probabilistic-classifiers-for-improved-transformer]
Text. Filename: Aizpurua_etal_TSMCS2018_Uncertainty_aware_fusion_of_probabilistic_classifiers_for_improved_transformer.pdf
Accepted Author Manuscript

Download (1MB)| Preview


Transformers are critical assets for the reliable operation of the power grid. Transformers may fail in service if monitoring models do not identify degraded conditions in time. Dissolved gas analysis (DGA) focuses on the examination of dissolved gasses in transformer oil to diagnose the state of a transformer. Fusion of black-box (BB) classifiers, also known as an ensemble of diagnostics models, have been used to improve the accuracy of diagnostics models across many fields. When independent classifiers diagnose the same fault, this method can increase the veracity of the diagnostics. However, if these methods give conflicting results, it is not always clear which model is most accurate due to their BB nature. In this context, the use of white-box (WB) models can help resolve conflicted samples effectively by incorporating uncertainty information and improve the classification accuracy. This paper presents an uncertainty-aware fusion method to combine BB and WB diagnostics methods. The effectiveness of the proposed approach is validated using two publicly available DGA datasets.


Aizpurua, Jose Ignacio ORCID logoORCID: https://orcid.org/0000-0002-8653-6011, Catterson, Victoria M. ORCID logoORCID: https://orcid.org/0000-0003-3455-803X, Stewart, Brian G., McArthur, Stephen D. J. ORCID logoORCID: https://orcid.org/0000-0003-1312-8874, Lambert, Brandon and Cross, James G.;