K-NN and Mean-Shift Algorithm Applied in Fault Diagnosis in Power Transformers by DGA

Power transformers submerged in oil are very important electrical equipment in the operation of an electrical system, they fulfill the essential role of transforming a level of voltage and current to the requirements of the User / Operator of the Electrical System; A significant aspect in the operation of an electrical system is the economic factor, since a contingency in the power transformer with service interruption can lead to considerable economic losses for the agents involved. In this work of investigation, is developed and analyzed a methodology for diagnose faults in power transformers by application of K-NN classifier with weighted classification distance, the training sample is considered the real data added to the densified data by mean shift algorithm (on the DGA sample), the performance recorded in the validation process is 97.73%.