EUVIMEDEUVIMED entdecken
Crossref

An Optimized Hybrid Ensemble Framework for Accurate and Scalable Transformer Fault Diagnosis Using Real-World DGA Data

Parineeta Ritesh Kelkar, Sharanabasava Inamadar

International Journal of Drug Delivery Technology · 11.06.2026 · 16 (52s)

DOI: 10.25258/ijddt.16.52s.7 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 11.06.2026. Verfasst von Parineeta Ritesh Kelkar und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Conversion transformers between power and other forms are important to modern power systems, with fault diagnosis being fast and accurate in order to provide a high degree of operational reliability and avoid catastrophic failures. In the past, Dissolved Gas Analysis (DGA) has been widely employed to keep track of the state of transformers; the classical rule-based interpretation techniques are generally less accurate and poorly generalize in high complexity and noisy operating conditions. To alleviate such challenges, the current study proposes an easy and scalable hybrid machine learning model of transformer fault detection using real-world DGA data. The suggested framework combatively analyzes a wide range of classical machine learning models, Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting (GB), Decision Tree (DT), and Logistic Regression (LR) as well as hybrid configurations of ensembles, such as RF+GB and SVM based. An intensive preprocessing and assessment pipeline, such as feature scaling and cross-validation are adopted, so that robust and fair comparison between models is ensured. The proposed method is also computationally efficient and has high diagnostic accuracy, as compared to deep learning-based methods, and therefore it can be applied to real-time and resource-constrained settings. As shown by experimental results on a real-world dataset of transformer faults, ensemble-based models are more efficient than single classifiers, with the proposed RF+GB hybrid model attaining up to 95% test accuracy and a macro F1-score of 0.88. Detailed class-wise and confusion matrix analysis show improved classification with minority fault classes with a strong generalization with low overfitting. The findings also indicate the significance of feature scaling in SVM based models and resilience of tree-based ensemble approaches to variability of features. The most important contribution of this paper is that lightweight classical machine learning and hybrid ensemble methods can be used to diagnose the fault in high-performance transformers without involving computationally intensive deep learning models. The proposed architecture is a scalable, practicable and deployment-ready structure of. intelligent transformer monitoring systems that will give the research accuracy with. practical implementation.

Zitieren

APA

Kelkar, P. R., & Inamadar, S. (2026). An Optimized Hybrid Ensemble Framework for Accurate and Scalable Transformer Fault Diagnosis Using Real-World DGA Data. International Journal of Drug Delivery Technology, 16(52s). https://doi.org/10.25258/ijddt.16.52s.7

Vancouver

Kelkar PR, Inamadar S. An Optimized Hybrid Ensemble Framework for Accurate and Scalable Transformer Fault Diagnosis Using Real-World DGA Data. International Journal of Drug Delivery Technology. 2026;16(52s). doi:10.25258/ijddt.16.52s.7

BibTeX
@article{Kelkar2026,
  author = {Kelkar, Parineeta Ritesh and Inamadar, Sharanabasava},
  title = {An Optimized Hybrid Ensemble Framework for Accurate and Scalable Transformer Fault Diagnosis Using Real-World DGA Data},
  journal = {International Journal of Drug Delivery Technology},
  year = {2026},
  volume = {16},
  number = {52s},
  doi = {10.25258/ijddt.16.52s.7}
}
Kostenlos starten →