Souza, A. N. [UNESP]Ramos, C. C OGastaldello, D. S. [UNESP]Nakamura, R. Y M [UNESP]Papa, J. P. [UNESP]2014-05-272014-05-272012-10-01INES 2012 - IEEE 16th International Conference on Intelligent Engineering Systems, Proceedings, p. 209-212.http://hdl.handle.net/11449/73612In this paper we propose a fast and an accurate method for fault diagnosis in power transformers by means of Optimum-Path Forest (OPF) classifier. Since we applied Dissolved Gas Analysis (DGA), the samples have been labeled by IEEE/IEC standard, which was further analyzed by OPF and several other well known supervised pattern recognition techniques. The experiments have showed that OPF can achieve high recognition rates with low computational cost. © 2012 IEEE.209-212engComputational costsDissolved gas analysisOptimum-path forestsRecognition ratesSupervised pattern recognitionForestryPattern recognitionPower transformersExperimentationPattern RecognitionPower FactorTransformersTrabalho apresentado em evento10.1109/INES.2012.6249832Acesso aberto2-s2.0-848666465288212775960494686