DGA-based diagnosis of power transformers - IEC standard versus k-nearest neighbours

The purpose of this paper is to present the comparison of application results of IEC-based classifier and the k-NN classic method to chromatographic data obtained by measurements on power transformers. To a large extent, the data reflect the state of power transformer and allow one to reason about the presence of possible faults. The distribution of real learning data is not even approximately uniform and makes the partitioning of decision space difficult.