9 th INTERNATIONAL CONGRESS ON ENGINEERING, ARCHITECTURE AND DESIGN, İstanbul, Turkey, 14 May 2022, pp.1043-1050
Abstract: Transformers can be designed and manufactured in different powers and sizes. The physical
dimensions of transformers change with their power. Transformers may need to be designed in different
winding types when their power changes. During the design, the powers of the transformer and the
voltage limits are taken into account. However, leakage fluxes are not taken into account for each size
and power. Because leakage fluxes occur during use after transformer design. Therefore, the leakage
flux can be estimated through simulation. The accuracy of the leakage flux values determined by simulation can always be doubted. Leakage flux is defined as leakage reactance in magnetic circuits. In this
study, a SVM model that can classify and detect transformer winding types by taking into account the
leakage reactance calculation has been designed and implemented. As a result of the test, it was seen
that SVM predicted the winding type with 99.7% accuracy. It was concluded that the SVM method has
the potential to save time in the selection of winding type in the design of transformers.
Keywords: Transformers, Coil, SVM, Cylindrical, Symmetric, Accuracy