EN

Scientific Publications

Important: This page is frozen. New documents are now available in the digital repository  DSpace

Abstract

Dissolved Gas Analysis (DGA) in oil is one of the most common and reliable techniques for diagnosing and early detection of incipient faults in oil-immersed transformers. However, conventional methods often encounter conflicts and challenges in accurately interpreting DGA results. Addressing this, two novel graphical methods, Circle Method 1 and Circle Method 2, are proposed, represented by circles with movable axes at variable angles. These methods innovatively base their analysis on the convergence of samples with similar characteristics around their centers and the spacing of these centers for samples with differing characteristics. The primary purpose of these methods is to distinguish between different overlapping faults. To achieve this, advanced metaheuristic algorithms are employed to determine the optimal angles that yield the best distribution of samples. Utilizing insights from previous DGA methods and field experiences, initial fault zones within Circle Methods 1 and 2 are estimated, and precise boundaries between fault zones are established based on practical DGA data collected from multiple sources. Evaluation using IEC TC10 database of fault cases demonstrates that Circle Methods 1 and 2 exhibit superior diagnostic accuracy compared to existing methods, including the Mansour Pentagon, the Duval Pentagon, the Gouda Triangle, and the Cartesian Graphical Method. This innovation significantly enhances the interpretative capabilities of DGA, offering a more reliable tool for fault diagnosis in oil-immersed transformers.

BibTex

@article{uniusa5170,
    title={A Novel Graphical Method for Interpretating Dissolved Gases and Fault Diagnosis in Power Transformer Based on Dynamique Axes in Circular Form},
    author={Abdelmoumene Hechifa, Chouaib Labiod, Abdelaziz LAKEHAL, Arnaud Nanfak and Diaa-Eldin A Mansour},
    journal={IEEE Transactions on Power Delivery}
    year={2024},
    volume={39},
    number={6},
    pages={3186 - 3198},
    publisher={IEEE}
}