Cekli S.Uzunoglu C.P.Ugur M.2024-07-122024-07-1220121992-86452-s2.0-84867525557https://hdl.handle.net/20.500.12415/8287The insulation condition monitoring of a power transformer has an important role for insulating materials which are subjected to extensive breakdown stress. In this study, a test setup has been constructed in order to simulate real world breakdown characteristics of transformerboards which are widely used as the insulating material. During the service life transformerboards may display undesired surface discharge damage due to increased rated voltages, which reduces the lifetime of transformerboards. The probabilistic neural network is used to detect the surface degradation of a transformerboard by analyzing electrical and ultrasound discharge data obtained from the test setup. The principle component analysis is employed to eliminate the messy matrix and vector calculations of the probabilistic neural network operations. Results of the classification procedure are given. © 2005 - 2012 JATIT & LLS.eninfo:eu-repo/semantics/closedAccessPrinciple component analysisProbabilistic neural networkTransformerboardArticle71Q4142