7 th INTERNATIONAL CONGRESS ON ENGINEERING AND TECHNOLOGY MANAGEMENT, İstanbul, Turkey, 16 - 18 April 2022, pp.14-21
Small and medium power monophase transformers usually have cylindrical
and two layered form. Therefore, two-layer and cylindrical winding transformers
are widely used in the energy sector in industry. The transformer has a
magnetic circuit and losses due to magnetism are one of the most important
factors affecting the performance of the transformer. The magnetic losses of
the transformer are mainly caused by leakage fluxes. Leakage fluxes are
represented by the leakage induction coefficient and are calculated depending
on the structural dimensions of the transformer. The performance of such
transformers, which is widely used in the industry, directly affects the users
and therefore the consumers. Artificial intelligence algorithms that can
automatically detect or classify the leakage fluxes of transformers can
facilitate the work of designers. In this study, five different machine
learning (ML) models are applied to detect and classify the leakage flux of
cylindrical two-layer monophase transformers. As a result, it was observed that
ML methods gave successful results in determining the leakage induction
coefficient.