Análisis del comportamiento de las variables dieléctricas del sistema de aislamiento eléctrico de transformadores trifásicos de distribución, en función de sus características eléctricas y de operación.

Abstract

Faults in a transformer are caused by mechanical or electric troubles or thermal stress; one of the reasons is the insulation damage, this leads to loses in the electric system due to blackouts or replacement of the damaged equipment. To reduce the presence of these type of faults, routine tests are applied to transformers, the mere application of these tests doesn’t imply a treatment of the results that may enable an insight analysis of the representation of a transformer. The present work is composed by: the enforcement of a stratified sample according to the use conditions, level of power and primary voltage, sorted by a clustering algorithm “k-means” of a representative sample of three phase distribution transformers that belong to the EEQ Transformers Lab; the power factor test was applied by the DELTA 4000 equipment; the results obtained were used to the construction of a linear regression model, that enabled to establish the correlation between the dielectric variables and the characteristics of the equipment tested. The results found in the linear regression model were compared with the results found by a black box model, obtained with an artificial neural network by modeling the data set obtained in the tests