ISSN (Online) : 2456 - 0774

Email : ijasret@gmail.com

ISSN (Online) 2456 - 0774


Survey on Application of SVM and KNN to DuvalPentagon 1 for Transformer Oil Diagnosis


Abstract

Abstract Dissolved GasAnalysis (DGA) is a widely used technique to estimate the condition ofoil-immersed transformers. The experimental results of the level and the changein concentration of different combustible gases in the insulating oil is atrustworthy diagnostic tool which can be used as indicator of undesirableevents occurring inside the transformer, such as hot spots, electrical arcingor partial discharge. The objective of this paper is mainly to analyseavailable data from DGA, and investigate data that may be useful inquantitative modelling of the transformer as reliability. Depending upon thelocation of a transformer, its rating and the nature of its usage, somedissolved gas analysis is to be scheduled which will be appropriate for thattransformer. The more critical the unit is the more frequently it should besampled. Here they have used the Support Vector Machine (SVM) and the K-NearestNeighbour (KNN) algorithms for the diagnosis of oil transformer. From these twoalgorithms from method known as a dual pentagon 1 diagnosis for this theyadopted five classes namely PD, D1, D2, T1 & T2, and T3

Keywords: Dissolved Gas analysis, Support Vector Machine,K-Nearest Neighbour, Dual pentagon 1


Full Text PDF

IMPORTANT DATES

Submit Article (Vol. 3 Issue 10) 

B
efore 10th Nov. 2018 

Issue Publication   On 10th Nov. 2018