Condition assessment of power transformer using SVM based on DGA

Abstract

The possibility of power transformer failure increases over the time as the age and rate of utilization increases. Since internal faults specially are the main cause of these failures, there are many ways and methods used to predict incipient fault and thus preventing the power transformer from failing by monitoring its condition. In oil immersed transformers, the DGA is used as one of the well-established tool to predict incipient faults occurring inside the body of power transformer. With already in existence of more than 5 known methods of DGA fault interpretation; there is the chance that all may give different conditions/results for the same sample. Using a combination of more than one of the methods and Support Vector Machine will result in increased accuracy of the interpretation and so reduces the uncertainty of the transformer condition monitoring.